Add per-connection discovery that loads canonical named tools on demand while preserving the default full catalog. Bound discovery replies before changing visibility, preserve family exclusions and Evidence workflows, and reset visibility on HTTP reconnect.
Verified with focused protocol tests and real Codex CLI artifact investigations; token costs remain task-dependent, so the profile is opt-in.
* fix(cli): reject project-scoped Amp registration
Incur's mcp add parses --no-global but its Amp writer always writes the
global settings file, so a project-scoped Amp request exited 0 reporting
success while changing global config. Fail closed in the entrypoint before
Incur dispatches: project-scoped Amp registration now exits 1 with a
diagnostic and leaves global settings untouched. Global Amp registration
is unchanged.
Fixes#7535
* fix(cli): exempt discovery requests from the Amp project-scope guard
Discovery requests are not registrations: mcp add --no-global with a
discovery flag (--help, -h, --llms, --llms-full, --version, --schema) now
passes through to normal handling instead of exiting 1 with the scope
error, matching the discovery set mcp add itself exempts. Real
project-scoped Amp registrations are still rejected before Incur runs.
Also restores the executable bit on scripts/rea.mjs, and adds launcher
dispatch tests proving the guard rejects a real registration before the
CLI runtime loads and passes discovery requests through.
* fix(cli): share the mcp add discovery classification between guard and routing
The Amp project-scope guard kept its own copy of the mcp add discovery
flags, which could drift from the discovery routing in cli.ts.
classifyMcpAddDiscovery in cliBuiltinArguments.ts is now the single
owner: the guard and the cli.ts mcp add special-casing both consume it,
with bypass/version/schema precedence matching the routing. No behavior
change; covered by classification unit tests and the existing mcp add
discovery serve test.
* fix(cli): emit structured output for Amp project-scope refusal
The Amp project-scope guard wrote a plain message to stderr and left
stdout empty when the caller asked for a structured document (json,
jsonl, yaml), contradicting the failure framing documented for the
output argument. Structured formats now render the refusal through the
same pre-execution error renderer the neighboring output-argument
branch uses, with code AMP_PROJECT_SCOPE_UNSUPPORTED on stdout; every
other format keeps the byte-identical plain stderr message. The guard
still exits 1 before dist/cli-runtime/cli.js is imported.
Co-authored-by: Saim Shafique <128062316+sx4im@users.noreply.github.com>
* test(cli): cover Amp project-scope guard through the launcher
The guard tests only called isProjectScopedAmpRegistration, so they
still passed with its call in scripts/rea.mjs removed. Add a launcher
test with an isolated home and working directory that checks the exit
code, the diagnostic, the unchanged global settings, and the absence
of project files for both explicit Amp and default-agent requests.
Co-authored-by: Saim Shafique <128062316+sx4im@users.noreply.github.com>
* test(cli): assert Amp project-scope refusal by error code
The --no-global refusal tests locked the full English sentence. Structured output now asserts AMP_PROJECT_SCOPE_UNSUPPORTED and a message field.
* fix(cli): share normalized mcp add agent selection with the scope guard
The project-scope guard scanned raw arguments and only recognized the spaced --agent form, while dispatch expanded --agent= equals forms first: --agent=claude-code was wrongly refused as default Amp, and --agent claude-code --agent=amp slipped Amp past --no-global. The equals expansion and agent scan now live once in cliBuiltinArguments.ts and both callers consume them, so guard and dispatch read the same selection.
* ci(environment): pin the real agentenv lane to agentenv-framework 0.9.1431
The real-agentenv lane, its identity check and the docs pinned
agentenv-framework 0.9.1334, 97 releases behind the current 0.9.1431.
Recorded verification notes keep the versions they ran with; the website
catalogs change only the placeholder comment that shows the version.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* docs(environment): describe 0.9.1431's pulled infra images; keep the site on the release
0.9.1431 pulls its gateway and service-db images from ghcr.io by the
digests its release records and builds one only when that pull fails,
so the lane docs, the export guide and the verifier's comments now say
so, and the mirror step fetches only the bases those builds name. The
public site describes the released package, which still runs 0.9.1334,
so its guide and evidence page keep that version until a release ships
the new pin.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
---------
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
* feat(native): report binary hardening for ELF, PE and Mach-O
Documents inspect_binary_hardening and its CLI equivalent, including
per-property states, supporting evidence, read budgets, and one report
per Mach-O slice. Explains that missing protection imports remain
unknown, signature presence does not establish validity, and hardened
runtime requires inspect_signature. Adds hardening links throughout the
documentation. The supplied diff also removes the advanced CodeQL
workflow and configuration, along with snapshot-binding mismatch
guidance.
* fix(native): validate declared payloads and dynamic tables before hardening rows
Run the shared range checks for ELF, read imports from the dynamic tables
rather than section metadata, and keep an unresolved symbol count apart from
an absent table. Check every Mach-O command before building rows and retain
only the offsets the rows need.
Make the verifier's readelf output wide, check the emitted format before
parsing Mach-O, and place the sources before the libraries in the link.
* fix(native): share Mach-O segment admission with hardening reads
A segment with a section count its cmdsize cannot hold passed the hardening
reader and could enable restrict_segment. Move the size, section count and
file range checks to a shared admission that readMachoImage and the hardening
reader both use, and pin the ELF load-size case with a regression test.
* fix(hardening): reject Mach-O section payloads past the slice and empty ELF interpreters
Segment admission never checked each section's payload range, so a
file-backed section past the slice still produced rows. Decode sections with
the section header owner and mark the slice malformed when a payload leaves
it; zero-fill and empty sections own no file bytes and are exempt.
admitInterp accepted a NUL first byte in a longer extent, which counted an
empty interpreter as PIE evidence. It now requires a nonempty path.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01KuWHGvzJkrhCqjtEHUwqyH
* fix(native): read CFG instrumentation from decoded GuardFlags
Use the IMAGE_GUARD_CF_INSTRUMENTED flag that guardFlagRecord already
decodes instead of masking the raw GuardFlags word a second time.
* fix(native): reject ELF table spans a second PT_LOAD remaps
The hardening reader mapped DT_STRTAB, DT_SYMTAB and hash tables through
the one PT_LOAD containing the table's first address. A second PT_LOAD
starting inside the table could map the rest of it from other file bytes
and the reader still reported parsed import evidence.
Cap each mapping at the next PT_LOAD that starts inside it, reject a
requested range that crosses that boundary, and stop hash reads at the
same limit, matching the conflicting-mapping refusal in readElfSymbols.
* fix(native): dedupe ELF note tables only on shared boundaries
The hardening reader dropped any note table whose bytes another table
already covered. A covering SHT_NOTE can hold a non-GNU note whose
descriptor is itself a complete GNU property note, while PT_NOTE names
that inner note directly. Walking only the outer stream missed the GNU
marker and reported CET disabled.
Drop an alias only when it starts and ends on boundaries of admitted
tables, which are note boundaries of the walked streams. Any other
covered alias leaves the feature rows unknown, as a partial overlap
already does.
* fix(native): admit every Mach-O symbol record after a canary match
The Mach-O stack protector scan returned as soon as it found an
undefined ___stack_chk_fail, so a later undefined symbol whose n_strx
lay outside the string table was never checked. The same damaged table
parsed or failed as malformed depending only on record order.
Keep the match and finish the bounded record walk. Records after the
match are only checked against the string table bounds, so their names
are not read or charged to the scan budget. The ELF reader already
admits its whole symbol table before matching names.
* fix(native): accept ELF load aliases with the same file translation
The hardening reader treated every PT_LOAD that starts inside a table,
or that also contains the table's address, as a conflicting mapping.
A load with the same address-to-file translation names the same file
bytes, so tables it overlaps still have one interpretation, and the
canonical ElfLoadMap already merges such spans.
Treat equivalent loads as one interpretation. A load with a different
translation, or the zero-filled tail of an equivalent load, still ends
the usable range, so a remapped table is still rejected.
* fix(native): keep same-span ELF note tables with different alignments unknown
A section and a program header that share endpoints were treated as one
note stream even when their effective note alignments differed. Walking the
same bytes at 4 and 8 can frame different notes, so dropping the second table
could hide a GNU property word and report CET or BTI as disabled.
Require matching effective alignment at both endpoints, and no alignment
change inside the covered span, before treating a table as an alias.
Otherwise mark the feature rows unknown. The fixture's .note.gnu.property
section now declares the class's note alignment, as toolchains emit it.
* refactor(native): read LC_SYMTAB through the shared fixed-word decoder
Move the Mach-O fixed uint32 command layouts out of MachoCommandNames
into domain/apple/machoLoadCommands so there is one owner of those
offsets. MachoCommandNames and the hardening reader now both decode
through it; the hardening reader keeps its own bounds and import policy.
* fix(native): skip Mach-O common definitions in canary imports
A common definition sets N_UNDF|N_EXT like an import but stores its
allocation size in n_value. An object holding only
`int __stack_chk_fail;` therefore reported stack_protector: enabled.
Require a zero n_value before an undefined external counts as an
import, for both nlist widths.
* refactor(native): read linkedit data commands through the shared fixed-word decoder
readMachoLinkeditDataCommand reproduced the dataoff/datasize offsets that
FIXED_WORD_FIELDS already owns. Take the command and decode through
decodeMachoFixedWordFields, as readMachoSymtabCommand does, keeping the
truncation and endianness behavior.
* fix(native): skip empty SHT_NOTE sections in ELF hardening
A zero-sized SHT_NOTE owns no bytes, so the section directory marks it
not file-backed and the hardening reader rejected the whole image as
malformed, while readElfImage accepted the same bytes. Skip empty note
sections before the file-backing check; they have nothing to walk.
* fix(scripts): run hardening verifier commands through the shared owner
The binary hardening verifier started its probes and compiler drivers with
raw spawnSync, which bypassed the src/process/ supervision that AGENTS.md
requires. Every launch now goes through createVerifierCommands, and the
temporary root is removed only after the owner closes cleanly.
* fix(scripts): clean up hardening verifier commands on Ctrl-C
SIGINT or SIGTERM ended Node before withVerifierCleanup ran, so detached
compiler trees and the temporary root could outlive the verifier. The
verifier now turns those signals into an abort that every owned command
receives, so cleanup closes the owner and removes the root before exit.
* fix(scripts): fail the hardening verifier when a signal lands outside a command
Only the owned commands read the interruption signal, so a SIGINT or
SIGTERM during the final cleanup or between commands was recorded and
then dropped, and the verifier exited 0. After cleanup settles, the
verifier now throws the interruption reason. A failure the operation or
its cleanup already raised still leaves first and is not replaced.
* fix(native): fail Mach-O slices with a malformed encryption command
The hardening reader turned every malformed status from the shared
encryption reader into an unknown `encrypted` row, so a short
LC_ENCRYPTION_INFO_64 or an encrypted range past the slice still left the
slice parsed with a full report.
Only two conflicting encryption commands stay an unknown row now. Any other
encryption defect fails the slice like other structural command failures.
* fix(native): ignore STB_LOCAL undefined ELF symbols as imports
A local undefined __stack_chk_fail or __memcpy_chk is never resolved
against another object, so it does not establish an external import.
Skip STB_LOCAL bindings when reading protector signs, and give the
fixture's positive symbols STB_GLOBAL binding.
---------
Co-authored-by: Matt Van Horn <455140+mvanhorn@users.noreply.github.com>
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
* fix(mcp): report target after failed open
* test(session): cover failed open receipt ownership
* test(session): verify restored target receipt
* test(mcp): retain target on snapshot mismatch
* test(mcp): retain oversized open failure receipts
* test(cli): avoid continuing session receipt in one-shot errors
* fix(mcp): report provider after failed switch
* test(ghidra): verify failed open preserves annotations
* test(mcp): join queued receipt request before cleanup
* test(session): assert open failure receipts in lifecycle suites
* test(session): keep open-failure receipt checks within function limits
Move the provider-collision case and snapshot fixture setup out of the
functions that grew when both failure receipts were asserted.
---------
Co-authored-by: Cursor Agent <cursoragent@cursor.com>
* feat(desktop): capture a Linux desktop application in use
Add `capture-desktop-application`, a CLI-only, Linux and X11 capture that
launches an installed desktop application, watches it while it is used and
writes capture Evidence plus SHA-256 addressed content:
- inventory: desktop entry, launch command, executable, package and its
apt sources (or snap, flatpak, AppImage), Electron markers and candidate
state directories;
- session: the launched process tree, client windows from xprop with the
first application window time and a window match, and files held open;
- state: a stat walk of the home directory before launch, at checkpoints
read from stdin, at stop and after exit, with created, modified, deleted
and retyped paths and the content of changed files and state directories.
It refuses non-Linux hosts, missing DISPLAY or xprop, a non-empty output
folder and an application whose window is already open, and signals only
processes it started.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* fix(desktop): own launched processes, confirm walk coverage and keep cancelled captures
- Launch through the owned provider path with a run token; release with
SIGTERM, then the token-verified group cleanup, after every session.
Processes without the token are never signalled; each records whether it
carries it, and the session records whether cleanup was verified.
- Report creations and deletions only where a walk listed the nearest
existing directory; count the rest as unconfirmed.
- Retain a file only when it is still the walked version, through owned
no-follow reads; charge the budget only for new content; create files
exclusively in a private output folder owned by the caller.
- Walk content roots outside the home directory, compare paths in their
physical spelling, and run executables from the caller's directory.
- Refuse when the window list cannot be read, re-evaluate window ownership
each sample, handle one queued request per sample with a bounded stdin
queue, and write the capture with end reason `cancelled` on interruption.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* fix(desktop): keep owned cleanup failures and fresh digests
- Allow a capture cancelled before any requested checkpoint to hold only the
baseline.
- Retry an incomplete stop once; if cleanup is still unverified, write the
capture and fail with the run token and process group as cleanup
resources.
- Classify path changes after a walk (ENOTDIR and other selected-file
errors) as changed_since_walk instead of failing the capture.
- Treat a non-directory ancestor as proof that nothing existed beneath it.
- Report a before digest only for the file version it was read from.
- Derive stderr_truncated from stderr's own byte counts.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* wip(desktop): state view over a capture folder
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* fix(desktop): keep failed file-handle closes owned for retry
Content reads pass retainFailedClose, so a handle whose close fails stays
owned by the content store; the capture retries those closes after it is
written and fails with the output folder as the cleanup resource if one
still cannot be closed.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* feat(desktop): read a captured application's state
Add inspect-desktop-application-state, a CLI command that reads a capture
folder into the application's state: every retained or changed file with
its role and reasons, a typed JSON view of each version (SQLite through the
shared worker, dconf through a private dconf dump, JSON, XML, key files,
text) and the kinds of change between versions (lines, values, keys, rows).
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* fix(desktop): report process and handle cleanup failures together
When both the launched processes and a retained file's handle remain
unresolved after the capture is written, one cleanup error now lists both
reasons and both resource sets.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* feat(desktop): export a captured application as an agentenv computer app
Add export-desktop-application, which writes an agentenv bundle with one
computer_app environment from a capture folder and its state view: the
apt package, a boot launch matched on the captured window class with a
window ready check, owned paths from the state roles, a seed of the
baseline files, data hooks that call agentenv's add/get/reset helpers
with a typed JSON view on get, a skill, a README and a deploy smoke task.
The desktop CLI commands now load their services lazily.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* fix(desktop): keep state views faithful to the capture
- Read key files, XML attributes, dconf groups and SQLite tables into
dictionaries without inherited properties, so captured keys such as
__proto__ stay data.
- Version databases with their write-ahead log as of each checkpoint, and
stage only digest-verified bytes for the SQLite and dconf readers.
- Detect the format per version, report format changes, keep JSON null and
empty containers as values, keep XML text and preserved whitespace, and
mark truncated views (including XML depth) and comparisons of them.
- Number versions and changes by checkpoint index, decide baseline
presence from the captured type, and keep tagged SQLite integers, text
bytes and blobs distinct.
- Carry the capture's limitations and omitted content forward, match state
directories by their home path, index changes once per capture, keep
failed handle closes owned, and finish the CLI through the shared owner.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* fix(desktop): keep coverage, logs and handles honest in the state view
- Read coverage flags only from formats whose readers write them, so JSON
fields named truncated stay data.
- Fail a SQLite version whose write-ahead log existed but was not retained
or does not match its digest, instead of reading the database alone.
- Close retained content on every exit, including cancellation, and report
a close that still fails together with the primary failure.
- Finish the CLI with the canonical operation name.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* fix(desktop): export only what the computer can install and seed
- Refuse packages not from the Ubuntu 24.04 archive, since the computer
installs by name from that archive.
- Start the launch command in the captured working directory.
- Respell the home path only in files the state view read as text.
- Report files that were not retained instead of dropping them silently.
- Load captured dconf keys for apps that own no files.
- Keep home-level files as owned paths; rename reserved Python class names.
- Mark truncated SQLite and text views in the class's get view, and read
file types the way the state view does.
- Release the output tree through a resource scope, and read seeds through
an owner closed on every exit.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* fix(desktop): keep staged folders owned, symlinked state roots and XML values
- Stage database bytes in a private runtime root released through a
resource scope, so a failed removal keeps its owner and is reported with
the read's outcome.
- Report cleanup facts and owners from desktop failures instead of
wrapping them as a plain adapter error.
- Record the state directories as walked, in their physical spelling, and
match roles against both spellings, so files in a symlinked state
directory are not mistaken for selected documents.
- Keep whitespace-only XML leaf values; drop only formatting between
elements.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* fix(desktop): read XML in the exported app the way the state view does
Keep whitespace-only leaf values and xml:space="preserve" text in the
generated get view; drop only formatting between elements.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* fix(desktop): release retained handles with their owners, read older captures
- Release retained-content handles whose close failed through a resource
scope on every exit; a release that fails again reports the exact handle
owners. A read that finished keeps its bytes when only its close failed.
- Accept captures written before state directories were recorded, and say
in the state view's limitations that symlinked state directories may be
misclassed for them.
- Compute each XML element's whitespace context once instead of scanning
siblings for every text run.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* fix(desktop): boot without the working folder, bound XML, accept mirrors
- Try the captured working directory only when an argument is relative,
and start the app even when that folder is missing.
- Cut the generated XML view at REA's node and depth limits, and keep one
unreadable file from failing data/get.
- Accept Ubuntu 24.04 mirrors, not only the canonical hosts; PPAs and
other repositories are still refused.
- Read seeds through the retained-content owner's new signature.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* fix(desktop): state view review round
- Stage a retained -journal beside its database so the SQLite worker
accepts an inactive journal and refuses a hot one; a -wal or -journal
that existed but was not retained (including one omitted at the
baseline) makes the version unreadable.
- Keep tables that read when another fails, and let cancellation and
unfinished cleanup propagate instead of becoming view errors.
- Read text as a key file only when every line parses as one.
- Rank a file named in a window title or selected by the caller as a
document even when the application holds it open; match titles on
file-name boundaries; leave the display name out of dconf matching.
- Bound JSON and key file views (4 MiB, 500 levels, exact numbers) and
report too-large, missing and altered retained content separately.
- Add no version or change while a database stays absent.
- Read staged databases in place instead of copying them again.
- Give Evidence the same limitations as the result.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* fix(desktop): export review round
- Load captured dconf keys without D-Bus: install dconf-cli, compile the
existing and captured key files into the user database, and warn on
failure instead of stopping the server.
- Start relative-argument launches in the captured working directory,
the home included, since the framework launches from the app folder;
respell the home itself and option values under it.
- Read views in a thread, skip files over 64 MiB, align the generated
readers with the state view's, and close SQLite connections.
- Never own the home or a shared base folder; map a symlinked state
directory's files back to their home spelling.
- Report omitted files and unretained opened folders.
- Name the module so it cannot shadow another, and keep the class off
the module's own names; derive names the bundle check accepts.
- Write seeds as they are read, and remove an unpublished bundle on
failure.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* fix(desktop): capture review round
- Ask the launch process group and token carriers to exit before
killing them, on errors too; report incomplete cleanup with its facts.
- Count descendants of the launch as started by the capture even when
their environment cannot be read; treat a reused pid as a new process.
- Follow symlinked desktop entries; name a flatpak's state after its
application id; match flatpak windows by name only.
- Stop on end of input; refuse empty, duplicate and reserved checkpoint
names; require whitespace after the checkpoint keyword.
- Stay on each walk root's device, and never walk or retain `/` or a
directory containing the home; say so in the limitations.
- Refuse a window manager that does not publish _NET_CLIENT_LIST.
- Remove retained content when a capture fails before it is written.
- Skip reading a file larger than the remaining content budget.
- Record apt's archive origin for each package source.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* fix(desktop): read an unset client list as no windows when supported
xfwm4 leaves _NET_CLIENT_LIST unset while it manages no windows, so the
capture refused an ordinary desktop with nothing open. Read
_NET_SUPPORTED with the list: an unset list means no windows when the
window manager supports it, and the capture is refused only when it
does not. `xprop`'s "no such atom" output now reads as unset too.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* fix(desktop): accept packages by apt's archive origin
Accept a package source when apt reports its origin as Ubuntu for a
noble suite, so mirrors count and PPAs or other repositories that merely
have "ubuntu" in their URL do not. Captures made before origins were
recorded fall back to Canonical's own archive hosts. Seed reads report
why retained content could not be used.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* feat(desktop): capture and export apps from vendor apt repositories
A closed-source app installed from its vendor's apt repository, such as
Typora, can now be captured and exported as a computer app.
- The capture records each non-Ubuntu repository listing the installed
version (uri, suite, components from apt's sources files, one-line or
deb822) and the key that signs it: the signed-by key, or the trusted
key that gpgv shows verifies the repository's signed index. The key
is retained with the capture.
- The capture records the packages owning the libraries the executable
loads (DT_NEEDED with RUNPATH and the linker cache, through REA's ELF
readers), since a vendor package may not depend on them.
- The export declares a verified vendor repository as an AptSource with
its key, installs a vendor package's library packages, and adds
--no-sandbox to an Electron app's launch, since a container denies
Chromium's sandbox.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* fix(desktop): report a locked database at once in data/get
A database the running app holds locked made each data/get wait sqlite's
five-second busy timeout before reporting it. Record the Typora run in an
agentenv computer in the verification.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* perf(desktop): detect key files without splitting the whole text
Read lines one at a time and stop at the first that a key file view
would drop, instead of splitting a file of up to 64 MiB into arrays
before its format is known.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* fix(desktop): keep symlinked homes and state folders under the home in exports
- Respell a path under its state folder's logical spelling only when that
spelling is under the captured home, so a symlinked HOME keeps its
files; pair walked and logical state folders only when the capture
lists the same number, and match the innermost folder first.
- When two files map to one home path, keep the one present before the
capture and name the other in the limitations.
- Read key files in the generated data/get with the state view's rule,
checked against it by running the helper under python3.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* fix(desktop): end captures on zombies and keep cleanup owners
- A zombie or dead process no longer keeps the session open.
- Remove only the partial capture file this run created.
- Keep the exact owners of handles and processes whose cleanup failed on
the returned error, so a retry can release them; test the published
capture's cleanup-failure branch with mixed outcomes and retries.
- Release the application once even when the exit checkpoint fails.
- A cancellation before the capture is written removes what it retained.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* fix(desktop): vendor package review round
- Match apt sources through `apt-get indextargets`, which covers flat `./`
and `/` suites and gives each release file and sources entry.
- Accept a key whose signature verifies even when the release carries
another signature without a key here; keep only the verifying key from
a keyring holding several (OpenPGP packet splitter).
- Install the package from its repository's release (`name/<release>`),
declare one source per URI and suite, support only the capture host's
architecture, and refuse Ubuntu Pro repositories.
- List only library packages the computer can install; walk libraries only
for vendor installs, with RPATH inherited as ld.so does, normalized
paths, and DT_NEEDED bounded while reading.
- Put --no-sandbox after the program an env wrapper launches; keep the
gpgv folder's owner when its removal fails; respect --max-file-bytes for
keys; wait 0.25 s on a busy SQLite database in data/get.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* docs(desktop): keep the vendor spec's gpgv command on one line
A code span broken across lines left `<binary keyring>` for VitePress to
read as an unclosed HTML tag, so the docs build failed.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* feat(desktop): add a validated task to exported desktop applications
Each capture can now yield one agentenv task in its exported bundle. An
agent writes the task (a prompt plus declarative checks over the app's
data/get views); REA validates it and stays model-free.
- `describe-desktop-application-task` summarises a capture's net change
per file (role, format, views before and after) and drafts checks.
- Checks: file presence, lines present/absent/counted, JSON-pointer values,
SQLite rows and counts. REA keeps a task only if a check fails on the
seed and every check passes on the captured end state; a file that was
not retained at the baseline counts as absent, as the computer starts.
- `export-desktop-application --task` writes tasks/NAME.json (deploy,
agent, skills, recording, prompt, verifier), a compiled verify.py that
reads data/get, and an eval. A boundary test runs the verifier under
python3 and checks it agrees with REA's evaluator on every check kind.
- Docs, spec and skill guidance, with the solve-rate recipe.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* test(desktop): keep the seed test in its own block
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* docs(desktop): record the task verifier driven in an agentenv computer
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* docs(desktop): record the Typora task run in an agentenv computer
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* fix(desktop): keep task checks bounded, complete and attributable
- line_count counts lines holding a substring (contains) instead of
matching a regular expression, which could stall the verifier on
agent-written text and read differently in Python.
- value_equals of a whole object or array is undecided on a truncated
view.
- The draft keeps a database whose rows changed under an unchanged hash,
and a file whose truncated views match while it changed.
- A task file that cannot be read keeps the reader's detail, and a failed
cleanup keeps the read failure as its primary.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* fix(desktop): leave a file with one version out of the task draft
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* docs(desktop): update the task verification's test counts
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* docs(desktop): record the task solve rate with Claude Code and Sonnet 5.5
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* feat(desktop): capture and export apps installed from a downloaded .deb
capture-desktop-application --package-file reads the file's control fields
with dpkg-deb and records its SHA-256, size and dependencies when it holds the
installed package; the file itself is never kept. The library walk now also
runs for local installs.
export-desktop-application --package-url declares agentenv's
Download(url, sha256) for that file, its dependencies and library packages as
apt packages, and a run script that installs it with dpkg.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* fix(desktop): seed app-named dotfiles and respell UTF-16 seeds
The capture now treats dotfiles in the home named after the application
(.varicad-cfg.ini, .notesrc) as state roots, so a file the app only reads
during the capture is still seeded. The exported seeding respells the
captured home in UTF-16 text files with a byte-order mark, keeping the mark.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* fix(desktop): hold readiness until the app's own window shows
An exported app is ready only once a window of its class that the taskbar
lists is shown, so a splash screen of the class, which skips the taskbar, no
longer counts as the app being up.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* docs(desktop): record VariCAD captured, exported and run in an agentenv computer
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* fix(desktop): keep exported views strict JSON for infinite and inexact numbers
The exported data/get spells infinite and negative-zero SQLite reals as REA's
SQLite reader does ("Infinity", "-Infinity", "-0"), and refuses a JSON file
holding a number that reading would change (1e999, integers past 2^53, -0,
rounded fractions) with the state view's own error, instead of a view with
an infinity in it.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* fix(desktop): keep package-file facts tied to one file and URLs unechoed
- A --package-url refusal never repeats the URL, and credentials are checked
before the scheme.
- The package file's identity must hold between dpkg-deb's read and the hash.
- A failed close keeps its cleanup owner even when the capture is cancelled.
- Readiness also ignores windows typed as splash screens.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* fix(desktop): record database logs and keep state changes distinct
- A selected path that is not a directory brings its -wal and -journal as
content roots, so the capture records whether they exist and retains them.
- The state view treats a file as absent only under recorded content roots
and refuses a database whose log presence is unknown.
- JSON changes list containers too, so an array that becomes an object with
the same members is a change.
- Key file entries escape the key as a JSON pointer segment, so [a/b] c and
[a] b/c stay apart.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* fix(desktop): keep watching through unread samples and late workers
- An unreadable window list is unknown, not zero windows, and is named in
the limitations.
- Before ending the session, live run-token carriers no parent link reached
are tracked; an unreadable process table leaves the exit unknown.
- An env prefix's assignments find the program and name its XDG folders.
- flatpak run options that take a value are skipped with it, -- is honoured
and a /arch/branch suffix is dropped from the id.
- State directories include the entry's full id.
- Path options refuse empty values; roots count toward the entry limit.
- A capture file whose close failed stays owned and is retried on discard.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* fix(desktop): bound apt sources, keep older captures and cleanup owners
- Sources files parse into one stanza per line or deb822 paragraph with its
URIs and suites kept as lists, so no URI x suite product is built.
- Repository architectures and release are optional, so captures written
before they were recorded parse; the export treats them as unrestricted
and unknown.
- Library sources are matched against every index of a declared
repository's URI, suite and components.
- Two failures side by side keep both causes and both cleanup owners, and a
host file whose close fails reports a lease on its handle.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* fix(desktop): match escaped key file entries and suggest containers through members
- The draft's key file lookup uses the state view's escaped `group/key`, so a
key holding `/` or `~` still gets its suggestion.
- JSON and XML changes now list containers too; the draft suggests an added
or removed container through its listed members, and a container only
when none of its members is listed.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* fix(desktop): read relative PATH entries from the launch folder
A program, or an empty or relative PATH entry, is resolved from the folder the
program starts in, as the launch does, so `env PATH=. app` with the entry's
Path still finds the app.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* perf(desktop): compare JSON views on one shared path
JSON and XML changes are found by walking both trees with one shared pointer
path, building a pointer only for a listed value, so memory follows depth
rather than every container's full path. A pointer over 4096 characters is
counted but not listed. A 256-level document with 4000-character keys went
from about 251 MiB of heap to under 1 MiB.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* fix(desktop): keep JSON Lines as text in the exported reader
The exported reader parses the whole text before an inexact number makes it
JSON with an error, so JSON Lines with an integer above 2^53 stay text as in
REA. Over-long integer spellings are unsafe without converting them. The
guide no longer claims large integers keep their value.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* fix(desktop): keep the state-path help free of dash-led words
The CLI alias check reads `-wal` and `-journal` in help text as short flags,
so the help names the write-ahead log and rollback journal instead.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* fix(desktop): never record or export apt source credentials
- Source URIs lose any user name and password as they are parsed, the way apt
splits them, and so do index-target URIs and descriptions and the sources
apt-cache policy lists, so matching works whether or not apt prints them.
- A repository whose entry held credentials is marked
`uri_credentials_removed`, and the export refuses it, or an older capture
whose URI still holds them, with a reason that names no URL.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* fix(desktop): look programs up without normalizing their paths
Program paths and PATH entries are joined rather than resolved, so `..` after
a symbolic link means what it means to the launch. The lookup moves to its own
module.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* fix(desktop): note inexact JSON numbers with a flag, not a list
The exported reader only needs to know whether any number was inexact, so it
keeps a boolean instead of every rejected spelling.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* fix(desktop): refuse older captures that still hold apt credentials
A capture written before credentials were removed counted the credentialed
repository's libraries as installable, so the export refuses it whole and asks
for a new capture rather than declaring packages no source offers. Recorded
policy sources are redacted in export messages too.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* test(desktop): keep the credential fixture on an example host
A login on a real host reads as a secret to the commit scanner, so the older
capture's fixture uses apt.example.com like the other credential tests.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* fix(desktop): tidy looked-up paths and search relative entries only for env
Empty and `.` segments are dropped from a looked-up program path, which keeps
`/snap/bin/` entries recognisable as snaps, while `..` is still left for the
kernel. Relative and empty PATH entries are searched only in the PATH an `env`
prefix sets, which travels with the exported launch; the host's own PATH does
not.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* fix(desktop): keep excluded and unwalked database logs unknown
The capture records the paths every walk left out, and the state view treats
a log as absent only under content roots the baseline walked completely and
outside those exclusions, so an excluded or unvisited log leaves the database
unread rather than read without it.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* fix(desktop): judge JSON numbers after duplicate keys settle
The exported reader marks an inexact number instead of flagging it during the
parse, then looks for marks in the parsed value, so a number a later duplicate
key overwrote no longer refuses the file, as in REA.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* fix(desktop): redact a repository's policy source when it names it
An older capture whose policy source matched no sources entry kept that text
whole, and the export named the repository by it; the export now names it
without any user name or password.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* fix(desktop): never signal the launch group on unknown membership
When the current read cannot observe a carrier's process group, the carrier
is signalled on its own instead of letting an older process-table row grant
a group-wide SIGTERM.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* fix(desktop): look programs up in the launch's order and refuse host-relative finds
The lookup searches PATH as the launch does, an empty entry meaning the
launch folder, so inspection and launch name the same program. A program found
only through a relative or empty entry of the host's own PATH is refused,
since the exported computer would not have it; an env prefix's PATH travels
with the command.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* fix(desktop): count unchecked paths and keep logs behind them unknown
Each walk counts the paths it could not stat and the directories on another
device it did not enter, and the checkpoint records it as `unchecked_paths`.
The state view treats a log as absent only when the baseline left nothing
unchecked, so a log whose stat failed stays unknown.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* perf(desktop): scan parsed JSON with one iterator per level
The check for marked numbers walks the parsed value with an iterator per
nesting level instead of copying each container's children; a million-entry
array went from about 15.7 MiB to 8.1 MiB of peak traced memory.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* fix(desktop): start a relative program from its captured folder
An exported launch whose program path is relative, such as `./app` from an
entry's Path, now changes to the captured working directory first, as it
already did for relative arguments.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
---------
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
* fix(mcp): guide retained evidence input recovery
* style(mcp): format catalog and server types
* test(mcp): verify HTTP capacity recovery guidance
* test(mcp): model catalog recorder imports in cleanup fixtures
record-mcp-catalog.mjs loads tool contracts and the Evidence-input
contract before it registers cleanup owners. The cleanup fixture still
sentineled only the previous imports, so those loads failed before any
owner wrote cleanup.jsonl. Track the script's read() calls and replace
the new modules with the empty contract list this fault injection uses.
---------
Co-authored-by: Cursor Agent <cursoragent@cursor.com>
* feat(native): page retained feature traces
* fix(snapshot): retain linked source evidence for replay
* test(snapshot): verify linked foreign evidence stays scoped
* docs(mcp): clarify imported residual questions
* docs(mcp): guide trace reference classification
* fix(cli): report result sources when Evidence publication fails
A trace keeps linked sources on the result. Use those records for the
undelivered source closure as well as the bundle, so a refused output
does not report them as not captured.
* ci: rerun pull request checks
The merge of origin/main did not start the CI workflow for this head.
---------
Co-authored-by: Cursor Agent <cursoragent@cursor.com>
* fix(native): retain value trace source evidence
* fix(native): build trace error evidence only on failure
* test(mcp): expect unowned value trace refusal
* test(mcp): avoid freezing trace error wording
* test(native): keep value-trace stop coverage with source evidence
Main's stop and continuation cases now meet the trace's evidence-owner
precondition: direct calls supply a source id, and a completed MCP trace
runs in a session that can retain it.
---------
Co-authored-by: Cursor Agent <cursoragent@cursor.com>
* fix(flutter): read unlabeled Dart Version::String() from engines
Release libflutter.so embeds Dart's Version::String() without the
standalone VM's "Dart VM version:" label, so identify_flutter_build left
dart_version null on real APKs. Accept that channel-bearing form beside
the labeled scanner path.
* fix(flutter): qualify unlabeled version candidates
---------
Co-authored-by: morluto <76467478+morluto@users.noreply.github.com>
* feat(binary-layout): admit ELF64 AArch64 and name the refused host
inspect_binary_layout refused every AArch64 ELF, and on a non-x64 host the
refusal read "supports Linux x64 only", which looked like a target refusal
(#8709 ran on arm64 WSL and suspected an e_machine check).
The pwntools adapter now admits little-endian ELF64 AArch64 EXEC/DYN/REL
objects alongside the existing profiles, pwntools must report aarch64/64/
little, and the Evidence subject names arm64. The host refusal now names the
detected platform/arch and says the target was not read; the host check
itself is unchanged.
verify:binary:layout cross-compiles an AArch64 executable and relocatable
object (never executed) and checks sections, symbols and relocations
against GNU readelf, decoding ELF64 r_info as a 32-bit type under a 32-bit
symbol index. The real-binary-layout workflow installs the toolchain.
Refs #8709
* test(binary-layout): refuse RISC-V as the unsupported machine and assert the host refusal path
verify:binary:layout used e_machine 183 (AArch64) as its unsupported
target, which this branch now admits; use EM_RISCV (243) instead.
The unsupported-host provider test now asserts the refusal names this
host's platform/arch and says the target was not read, so reverting the
decoder to the old message fails it.
* fix(native): stop calling an off-screen window absent
The native UI helper looked up its selected window with
`.optionIncludingWindow`, which returned nothing while that window was off
screen. An existing window owned by the selected process was reported as
`window-mismatch`, before any permission check, so the verifier could not reach
its permission boundary. Select from all windows instead; absent and foreign
windows still fail.
The verifier's changed-bytes check also read `error.message`, but since
cc90771af a selection failure carries its helper code in `error.issues`, so the
check could not pass. Read the issue. Together these let the permissions lane
complete on main.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* test(native): own UI test processes from the moment they launch
The off-screen test registered cleanup only after parsing the ready record, so a
startup failure left `/usr/bin/open` and its app running. The lane test killed
only the verifier on a timeout, leaving its launcher and detached fixture alive
until they expired.
Register the off-screen owner immediately after the launch; the app inherits the
run token, so the sweep needs no ready record. In the lane test, every stand-in
records itself and leaves on one release the test's cleanup awaits.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
---------
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
* fix(debugger): let REA exit when a debugger could not be stopped
`rea debug-with-gdb` and `rea debug-with-rizin` printed cleanup_incomplete
and set exit status 1, but the retained child and its pipes kept the event
loop alive, so the process never exited. A stdio MCP server whose client
disconnected hung the same way.
closeOwnedDebuggerSessions is the owner both managers finish through. It now
unreferences retained sessions, as supervised providers already do after
incomplete cleanup, restores the references on retry, and names the retained
process IDs. The one-shot commands report those IDs, since no later command
can retry. Debugger ownership, signals and detach behavior are unchanged.
GDB cleanup requests also no longer hold the host until their 120 s command
deadline, because cleanup already bounds its own wait.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* fix(debugger): name retained debuggers when an MCP server shuts down
An MCP server that exits with a retained debugger logged only "REA provider
cleanup failed" and "could not close cleanly", so nothing named the process
left for the caller to check. The PIDs were on OwnedDebuggerCleanupError, but
each shutdown reporter read only the outer aggregate's message, and the debug
failure_cause description dropped the field too.
The failure type now lives in its own dependency-free module with one reader
that walks composed failures. The shutdown report, the server's onclose log,
the debug description and the one-shot CLI all use it, and the shutdown report
writes the same remediation the CLI returns.
The fake debuggers now model each request exactly, with real reply shapes, and
a request they do not model fails the test.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* test(debugger): compare retained-debugger guidance with its formatter
The new checks froze the wording "Check process ... and its target".
docs/testing.md asks guidance checks to compare with the owning formatter,
alongside independent checks of the facts. They now do, and each still proves
the process ID is named, because a formatter comparison alone would pass if the
formatter dropped it.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
---------
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
* fix(native): find procedures by the secondary names exact lookup accepts
search_procedures matched only each procedure's primary name, while exact
lookup also accepts Kuna inventory aliases (versioned-symbol and generated
sub_ names) and Ghidra symbols at an exact function entry. A search for a
name procedure_address resolves returned nothing, and trace_feature
silently missed the procedure.
Both providers now match the same names their exact lookup accepts. A
procedure still appears once under its primary name; the optional
matched_aliases field lists the secondary names that matched, so aliases
never become procedures of their own and list_procedures is unchanged.
The shared domain helper owns that row rule for Kuna and for the Ghidra
host-side Unicode folding pass. The Ghidra analysis profile records the
new search semantics so older snapshots cannot replay alias-blind results.
Fixes#8330
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* fix(ghidra): match leaf spellings of procedure names in search
procedure_address accepts each entry symbol's leaf or namespace-qualified
spelling, but procedure search tested only qualified names, so a regex
such as ^audit_alias$ missed a secondary symbol ns::audit_alias (and ^same$
missed a primary alpha::same) that exact lookup selects.
Lookup and search now share one bridge definition of the names that select
a procedure, so they cannot drift. Matching secondary symbols are still
reported by their qualified names; because a leaf ends its qualified
spelling, the host's Unicode literal pass stays exact over those names.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
---------
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
* fix: retain address context cleanup failures
* fix!: preserve address context input errors
BREAKING CHANGE: Strict clients that cached the previous full inspect_address_context output schema must refresh tools/list and accept the typed error field on unavailable facets.
* test: use session contract for address context
Bind typed service creation, observed availability and retryable cleanup in one composition owner. Reuse selected CLI/MCP factories and canonical contract groups without copying provider operation inventories.
Closes#8429. Refs #740.
Validated paired real application workflows, actual local Codex CLI Terra/high use, complete Evidence and Unknown checkpoint parity, compiler-backed import boundaries, SDK schemas and final real mixed routing. Automated review findings were fixed at their producing owners. Existing exact-base CodeQL alerts are documented on the PR under the owner-authorized unrelated-check exception; no global alerts or security rules were changed.
Assisted-by: Suika (gensokyo-skills:suika-scatter-gather-planning)
The DOS verifier compared a fresh CLI session with an MCP session that
its address-boundary helper had renamed and restored, so the entry's
symbol source differed (imported vs user_defined). Compare the CLI with
the pre-edit MCP window, then assert the restoration changes only the
symbol provenance.
Fixes#8189
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
swiftc rejects assembly sources as inputs, so verify:ghidra:swift failed
during fixture preparation before any provider check. Assemble both
assembly fixtures to objects with xcrun clang and link only objects.
Fixes#8188
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
* fix(ghidra): decide batch names independently of request order
The Ghidra bridge applied set_addresses_names in JSON member order, each
address in its own transaction, so a reordered but equivalent map could give
a shared name, or a name released by another selected address, to a
different address.
The bridge now resolves every destination (namespace and leaf name) against
the names held before the batch. Addresses that request the same destination,
or a name that another selected address holds, are rejected without applying;
the remaining edits are independent and apply as before. Both naming
operations report each failure's canonical address, reason and related
addresses. The adapter projects the bridge reply to the unchanged boolean
results and keeps the reply as raw_result only when an address failed.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
* fix(ghidra): report each contested batch name once
A batch conflict listed every other requester on each failure, so k
addresses competing for one name produced k(k-1) related addresses, and
the adapter retained the whole reply, booleans included, as raw_result.
The bridge now reports each contested destination once in a conflicts
table with its namespace, name, requesting and holding addresses, and
each conflict failure refers to it by index. Conflicts are ordered by
their first requester, so indexes do not depend on member order. Single
and batch naming share one planner and reply layout. The adapter
validates the table in linear time and keeps only failures and conflicts
in raw_result.
Co-Authored-By: Claude Opus 5.5 <noreply@anthropic.com>
---------
Co-authored-by: Claude Opus 5.5 <noreply@anthropic.com>
gensokyo-dev