_parse_stream_json() persisted raw tool_input/tool_response content into
ObservationEvents that grade() scores and generate_report() writes to
results/<skill>.md -- a report meant to be shared and reviewed.
--add-dir restricts the agent's additional accessible directory to the
sandbox (SANDBOX_BASE = /tmp/skill-comply-sandbox), but that doesn't stop
the agent's own tool calls (a Bash command using ~ expansion, a scenario
setup_commands entry referencing a dotfile) from emitting the operator's
home directory into tool_input/tool_response -- which then lands
verbatim, truncated but not sanitized, in the written report.
Adds _redact_home_path(), pure stdlib (Path.home()), applied to both
input_str and output_str before they're stored on the ObservationEvent.
Scoped deliberately to the home directory only -- grade() needs real
tool-call semantics for LLM-based compliance classification, so
truncating/stripping content the way a pure logging hook could isn't an
option here; only the operator-identifying path component needs to go.
New TestParseStreamJsonRedactsHomePath class in
skills/skill-comply/tests/test_runner.py (3 tests) -- full file now
10/10 passing, up from 7/7. Confirmed tests/test_invariant_runner.py (the
sandbox-execution security tests from #2149) still passes clean, 4/4.
Fixes#2730
The find -> sort -> read chain in both scripts used newline-delimited
records (plain read -r), so a skill directory name containing a
literal newline would be split across two records. Verified with a
directory literally named "evil\nskill": the old reader produced a
truncated "evil" fragment plus an orphan "skill/SKILL.md" fragment,
inflating the skill count and throwing awk/date errors on the garbage
paths.
Switch to -print0 / sort -z / read -r -d '' in both scripts so a path
is always read as a single record, regardless of its contents. Paths
under scan are untrusted input.
Addresses further CodeRabbit review feedback on PR #2640.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01FhDjpSfrbPpnpqT3CZBEX1
Following up on the -L fix: find -L can now traverse symlinks, but a
broken symlink target or an unreadable directory makes find skip that
entry and exit non-zero. Both scripts previously redirected find's
stderr to /dev/null and never checked its exit status, so a scan could
silently under-count skills with no indication anything was wrong.
Capture find's exit status and stderr in both scripts; on failure,
print a warning (with the underlying find error) to stderr while still
emitting the best-effort results for whatever was found. Verified with
a permission-denied skill directory: real BSD find exits 1 and reports
"Permission denied" on stderr, now surfaced as an explicit warning
instead of silently dropped.
Addresses CodeRabbit review feedback on PR #2640.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01FhDjpSfrbPpnpqT3CZBEX1
scan.sh and quick-diff.sh both used `find "$dir" -name "*.md" -type f`,
which missed symlinked skill directories (no -L) and miscounted any
non-skill markdown file sitting in a skills directory as a skill
(matched *.md instead of SKILL.md). Both call sites now use
`find -L "$dir" -name "SKILL.md" -type f`.
Repro (temp dir with 1 real skill, 1 symlinked skill, 1 stray .md file):
before: 2 skills found (real skill + the stray .md, symlinked skill invisible)
after: 2 skills found (real skill + symlinked skill, stray .md excluded)
Fixes#2598
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01FhDjpSfrbPpnpqT3CZBEX1
Extends scripts/ci/validate-skills.js to also scan docs/{locale}/skills/
translated mirrors, not just curated skills/. Adds detection for the
YAML defect classes from #2630 without a parser dependency: unquoted
values containing ": " (glued next key / dropped quoting), values
starting with the reserved '@'/'`' indicators, and missing frontmatter
blocks entirely (required only for docs mirrors; curated skills/ keeps
its existing tolerant behavior).
Greptile flagged that the env-access guard in
gateguard-env-documented.test.js could be bypassed via reflective reads
of process.env (Reflect.get/has/set/deleteProperty/defineProperty/
getOwnPropertyDescriptor/ownKeys), since none of the existing
UNSUPPORTED_ACCESS patterns matched that form.
Add a rule that rejects Reflect.get/has/set/deleteProperty/
defineProperty/getOwnPropertyDescriptor/ownKeys(process.env, ...) and
three self-check fixture cases (Reflect.get, Reflect.has,
Reflect.ownKeys) so the guard is pinned against silently missing them
again.
Negative control: commenting out only the new rule reproduces exactly
the reported gap (the 3 new fixture cases fail with "access guard
missed: Reflect.get, Reflect.has, Reflect.ownKeys"); restoring it goes
back to 10/10.
The documentation surface test scanned the hook's raw source with two
regexes. That had two holes, both confirmed against the shipped parser:
- a GATEGUARD_* name appearing only in a comment or a string was counted
as a real read, and
- destructured, aliased and computed reads were invisible, so an
undocumented knob added in one of those forms would pass silently.
Blank comments, string literals, template-literal text and regex literals
before scanning, so only real code contributes. Blanking preserves length,
so `process.env[...]` keys are located in the blanked code and read back
from the raw source at the same offset.
Rather than chase every possible access form with regexes, the supported
forms are now enforced: destructuring, aliasing, spreading, enumerating
and computed keys fail the guard with instructions to either keep the
convention or extend the parser. Six self-checks cover the blanker and
the guard, including a regex literal containing a slash.
Refs #2573
CodeRabbit review on #2611, all four findings:
- GATEGUARD_DISABLED sat in a table introduced as 'these do not disable the gate'. Moved to its own full-disable section with ECC_GATEGUARD, and corrected the accepted values against ECC_DISABLE_VALUES (0/false/off/disabled/disable - the earlier draft would have implied 'no' works, which it does not).
- Documented that a leading **/ compiles to .*/ and so needs a preceding separator: verified by reproducing the hook's glob->regex translation, **/tests/** matches /repo/tests/foo.js but not a bare relative tests/foo.js. Docs now say so and the example carries both forms. Matcher behaviour deliberately unchanged - widening it is a behaviour change, not a docs fix.
- Reverse-drift check now compares documented names against the parsed env reads instead of hookSource.includes(), so a name surviving only in a comment or error string no longer satisfies it.
- readGateguardEnvNames builds one Set from collected matches instead of mutating via Set#add, per the repo's no-in-place-mutation guideline.
GateGuard reads five GATEGUARD_* environment variables that were absent
from skills/gateguard/SKILL.md, so the only discoverable escape hatch was
ECC_GATEGUARD=off - disabling the load-bearing destructive-Bash gate
along with the noisy ones (#2573).
Documented, with defaults and exact accepted values read from the hook:
- GATEGUARD_BASH_ROUTINE_DISABLED (was undocumented everywhere)
- GATEGUARD_EXEMPT_GLOBS (previously only in a 2.1.0 release note)
- GATEGUARD_BASH_EXTRA_DESTRUCTIVE (was undocumented)
- GATEGUARD_DISABLED (was undocumented)
- GATEGUARD_STATE_DIR (was undocumented; named in a runtime warning)
- GATEGUARD_FACT_FORCE_FULL_DENIALS (already documented; folded into the
same table for one lookup point)
Adds tests/ci/gateguard-env-documented.test.js, which asserts every
GATEGUARD_* variable the hook reads appears in the skill doc, and that the
doc names no variable the hook has stopped reading. That surface test is
what found the three knobs beyond the two the issue reported.
Docs and test only; no hook behaviour changes.
Refs #2573
The context signal always rendered "N% of <window> window", including when
the window size was the assumed 200k default rather than a detected value.
On a 1M session whose transcript carries no [1m] marker, that produced
lines like:
[StrategicCompact] Context ~194k tokens (97% of 200k window)
while actual usage was ~19%. The user compacts on a false alarm, loses
context, and the resulting quality drop reads as a model regression.
The gap is structural: the context threshold defaults to 80% of the
window (160k on 200k), so the signal fires precisely in the 160k-200k
band where the size cannot be determined — above 200k the observed-tokens
fallback correctly infers 1M, and below 160k nothing fires.
Model id alone cannot close this. A tier may ship both a 200k and a 1M
variant under one id, so neither the known-family table nor a new entry
can distinguish them, and the transcript records no window field.
So stop asserting what isn't known: resolveContextWindow() now reports
whether the size was detected (env override, [1m] marker, known family,
or observed tokens > 200k) or assumed, and the hook omits the percentage
and window label when it was assumed. The token count, threshold, and
firing behaviour are unchanged.
resolveContextWindowTokens() keeps its existing signature and semantics.
Note: 3 pre-existing failures in tests/hooks/suggest-compact.test.js
reproduce identically on unmodified main and are untouched here.
opencode's legacy plugin loader (getLegacyPlugins) iterates every module
export and throws 'Plugin export is not a function' if any export is not
a plugin function. The bundle exported VERSION (string) and metadata
(object) alongside the plugin, breaking plugin loading. Export only the
plugin function so opencode can load ecc-universal.
Following the hermes/openclaw (#2433) and kimi (#2441) adapter recipe.
What's included (adal-project adapter, project kind, ./.adal root, same
shape as kimi-project/joycode-project):
- scripts/lib/install-targets/adal-project.js — 10-line project-kind
adapter targeting ./.adal
- Registry + helpers platform-ownership wiring
- adal target on the 5 shared modules (rules-core, agents-core,
commands-core, platform-configs, workflow-quality) +
SUPPORTED_INSTALL_TARGETS + legacy-compat module
- .adal in platform-configs paths
- Both schema enums (install-modules, ecc-install-config), npm files
allowlist, installer help text, .adal/README.md stub
AdaL (adalagent.ai) is a terminal-based AI coding agent (by SylphAI)
built on AdalFlow, with native MCP support and project-scoped config
under ./.adal/ (skills, custom tools, memory) plus a root-level
AGENTS.md instructions file — matching the shape ECC already installs
into other AGENTS.md-based harnesses (Codex, OpenCode, Kimi).
Verified: full suite matches main's baseline (3334 passed, same
pre-existing failures unrelated to this change — OpenCode build/npm-pack
surface tests requiring build tooling not present in this sandbox);
catalog check passes (67 agents / 94 commands / 281 skills); dry-run
resolves Target: adal / Adapter: adal-project / root ./.adal with all 5
modules planned; doctor reports OK after a real install; uninstall
cleanly reverses all 458 operations.
Co-Authored-By: AdaL <adal@sylph.ai>
Keep the contributor's corrected Haiku and Fable/Mythos tiers, update the live Sonnet and Opus rows against the current official pricing contract, and mirror the same numeric changes in the Japanese and Chinese tables.
Keep the duplicate estimator and its tests deleted as landed through #2866, while carrying the contributor's pricing-table correction forward and refreshing the documented live model rows against the current official pricing contract.
Clarify the editor boundary, require deliberately bounded egress, qualify Jailbox as one verified reference pattern, and remove the duplicate AWS bulletin citation.
Remove the redundant JSON-RPC initialize fallback from the consolidated MCP health-check batch. A routed 404 already proves the endpoint is reachable, and the real authenticated MCP call remains authoritative. Avoiding the fallback also prevents a stalled GET plus stalled POST from consuming twice the configured hook timeout.
