8d58000fab
fix(harness): classify opencode terminal output and disable project config in all eval cells
...
The opencode review branch parsed output without classify_route_output,
so a terminal 529 left PROVIDER_OUTCOME=ok and skipped the required
same-route retry; run the shared classifier first in both review scripts.
The write-capable eval cell now also sets OPENCODE_DISABLE_PROJECT_CONFIG
so a fixture's .opencode/{plugins,agents} cannot contaminate the eval
before wrapPrompt applies.
Claude-Session: https://claude.ai/code/session_01EoF39fuXHnmWDVzkBADnzo
2026-08-31 13:33:02 -07:00
98485c9cea
fix(harness): disable opencode project config on read-only review routes
...
A reviewed repo's .opencode/{plugin,agent} directories load as trusted
runtime and can run host shell or override the global deny overlay, so
the read-only peer was a host RCE (reproduced live: a planted
.opencode/plugin ran shell despite edit/bash denied). Set
OPENCODE_DISABLE_PROJECT_CONFIG=1 on every review/POV route and the
eval-cell read-only arm; verified reads still work with it set.
Claude-Session: https://claude.ai/code/session_01EoF39fuXHnmWDVzkBADnzo
2026-08-31 13:00:20 -07:00
df2b1ad13a
fix(harness): fail closed on opencode review routes and scrub cell attestation
...
The read-only overlay now also denies webfetch and task (live-tested:
reads survive, fetch and delegation are gone; a wildcard deny was
rejected because it removes read tools too). Eval cells scrub every
inherited harness-attestation marker, not just CLAUDECODE, so a child
cell launched from Codex or Grok attests as itself.
Claude-Session: https://claude.ai/code/session_01EoF39fuXHnmWDVzkBADnzo
2026-08-31 12:51:22 -07:00
db8c61d126
fix(harness): enforce opencode read-only in the eval cell and admit it at the target gate
...
The eval cell's read-only arm only omitted --auto, which is not
read-only under OpenCode's permissive defaults; it now injects the same
deny overlay the peer adapters use. cross-model-execution's allowlist
sentence still excluded opencode while the route token admitted it.
Claude-Session: https://claude.ai/code/session_01EoF39fuXHnmWDVzkBADnzo
2026-08-31 12:30:54 -07:00
980f959024
feat(harness): add opencode as a named peer and work engine
...
Operators can set cross_model_peer and work_engine harness to opencode.
Review and POV run `opencode run` with OPENCODE_CONFIG_CONTENT denying
edit and bash. Work uses --auto. Output is parsed as OpenCode NDJSON
text events, not grok or Cursor envelopes.
2026-08-31 12:30:49 -07:00
98932f162e
fix(ce-work): verify worktree snapshot fidelity before isolated dispatch ( #1602 )
2026-08-31 11:56:44 -07:00
012496218f
fix(ce-work): out-of-repo units have no git-derived completion ( #1583 )
...
Co-authored-by: Trevin Chow <trevin@trevinchow.com >
2026-08-31 11:56:17 -07:00
fe02f790d3
fix(ce-code-review): isolate testing reviewer mutation work ( #1584 )
...
Co-authored-by: Trevin Chow <trevin@trevinchow.com >
2026-08-31 11:12:41 -07:00
1b170d26f0
fix(ce-babysit-pr): persist review invariant rounds across heads ( #1585 )
...
Co-authored-by: Trevin Chow <trevin@trevinchow.com >
2026-08-31 11:05:04 -07:00
42c6c709d7
fix(ce-work): require evidence before unavailable review fallback ( #1578 )
...
Co-authored-by: Trevin Chow <trevin@trevinchow.com >
2026-08-31 10:56:38 -07:00
c7e3c14a8e
fix(ce-commit): make the stage-and-commit example PowerShell-safe ( #1587 )
...
Co-authored-by: Trevin Chow <trevin@trevinchow.com >
2026-08-31 10:56:27 -07:00
c9c10f8c75
fix(codex): reconcile config-only plugin installs ( #1599 )
2026-08-31 00:38:32 -07:00
bc26f23a25
fix(ce-work): allow parallel subagent waves in a shared workspace ( #1598 )
2026-08-31 00:30:03 -07:00
5f5bc6b965
fix(ce-babysit-pr): keep stale-computation degrade off the settle clock ( #1568 )
...
Co-authored-by: Trevin Chow <trevin@trevinchow.com >
2026-08-30 21:01:15 -07:00
a79582c3a4
fix(ce-commit-push-pr): catch a PR opening that names the mechanism ( #1595 )
2026-08-30 20:58:18 -07:00
debd05a00f
fix(ce-commit-push-pr): let a PR opening carry what motivates it ( #1576 )
2026-08-30 20:27:15 -07:00
ba4468fa8f
fix(ce-plan): require a holdable Goal Capsule Objective ( #1592 )
2026-08-30 19:56:48 -07:00
0e758b60b3
fix(lfg): record unapplied review findings in the PR body, not tickets ( #1580 )
2026-08-29 15:58:18 -07:00
a1f601f171
fix(ce-setup): detect the retired Codex tool map ( #1577 )
2026-08-29 01:58:55 -07:00
f6c301cafc
fix(ce-code-review): take review criteria from a repo-owned standards file ( #1572 )
2026-08-28 14:34:52 -07:00
f76d3096a1
fix(guides): keep catalog out of the plugin skills tree ( #1571 )
...
Co-authored-by: Cursor Agent <cursoragent@cursor.com >
2026-08-28 16:15:16 -05:00
84bdf8c5a1
fix(ce-compound): check in-repo absolute path citations ( #1561 )
...
Co-authored-by: Cursor Agent <cursoragent@cursor.com >
2026-08-27 21:27:19 -07:00
5985d821b9
perf(tests): reuse seed git fixtures to cut suite wall time ( #1548 )
...
Co-authored-by: Cursor Agent <cursoragent@cursor.com >
2026-08-26 22:14:16 -07:00
f662cad495
docs(guides): move skill catalog from docs/skills to skills/guides ( #1551 )
...
Co-authored-by: Cursor Agent <cursoragent@cursor.com >
2026-08-26 19:37:39 -05:00
4e6738223d
fix(ce-compound-refresh): preserve supported guidance across regressions ( #1542 )
2026-08-25 15:33:38 -07:00
118633f104
fix(peer-job-runner): keep Windows reap from killing a recycled pid ( #1541 )
2026-08-25 08:19:50 -07:00
26bf5b1e37
fix(concepts): give CONCEPTS.md a retention lifecycle ( #1540 )
2026-08-24 23:48:31 -07:00
e322bdacd4
fix(ce-code-review): keep missing peer config from skipping the pass ( #1538 )
2026-08-24 16:05:45 -07:00
2ca1f81cd5
fix(skills): anchor the plan Objective above the component being changed ( #1535 )
2026-08-24 14:48:23 -07:00
b20c29d7aa
fix(cross-model): drop empty Codex usage files and unblock heartbeat teardown ( #1533 )
2026-08-24 12:37:15 -07:00
5ac4c01a42
fix(ce-code-review): collect async reviewer returns ( #1530 )
2026-08-24 00:19:33 -07:00
5d5b78431a
fix(ce-babysit-pr): defer to GitHub for trunk drift on managed stacks ( #1529 )
2026-08-24 00:17:31 -07:00
d05e3951cc
fix(lfg): invoke ce-code-review from the host catalog path ( #1527 )
2026-08-23 23:34:46 -07:00
180fcd21e1
fix(skills): match blocking questions from the current tool list ( #1526 )
2026-08-23 23:27:35 -07:00
f293ca43cb
docs(readme): make the README a front door, split detail into linked docs ( #1525 )
2026-08-23 23:08:27 -07:00
2e10508950
fix(cross-model): bound peer retries and raise review headroom ( #1519 )
2026-08-23 09:16:20 -07:00
2d77dee3da
fix(ce-doc-review): activate product-lens only on a product position ( #1517 )
2026-08-23 00:09:11 -07:00
3ce5855fab
fix(cross-model): attest Grok Build as a host and bind its native CLI ( #1516 )
2026-08-22 22:34:22 -07:00
6f6c5779d3
fix(skills): right-size ceremony for small work in ce-plan, ce-brainstorm, and ce-work ( #1514 )
2026-08-22 20:39:30 -07:00
dfeae09526
fix(ce-code-review): stop the reviewer template from embedding the diff twice ( #1512 )
2026-08-22 08:57:36 -07:00
f7efe88675
docs(agent-plugins): attribute the 8000-byte skill cap to its real owner ( #1511 )
2026-08-22 00:32:13 -07:00
28ac832285
fix(cross-model): accept provider-qualified codex model ids in cross_model_model ( #1501 )
2026-08-21 22:25:26 -07:00
9107007954
test(cross-model): give the subprocess-heavy route suites a 30s per-test ceiling ( #1510 )
2026-08-21 15:59:16 -07:00
bcd86d0449
fix(skills): load ce-plan and ce-work phases from references under the 8KB cap ( #1508 )
2026-08-21 15:36:14 -07:00
6dda2038b1
fix(ce-optimize): keep multi-objective wins and screen expensive runs ( #1506 )
2026-08-21 15:12:24 -07:00
66ccf579f8
fix(skills): remove obsolete dispatch context hook ( #1499 )
2026-08-20 23:30:13 -07:00
dedf301348
fix: honor active model-config keys in skill runs (surface + salience) ( #1489 )
...
Co-authored-by: Trevin Chow <trevin@trevinchow.com >
2026-08-20 20:28:11 -07:00
3c04954458
fix(cross-model): elevate peer start when host sandbox blocks provider network ( #1496 )
2026-08-20 17:59:09 -07:00
12d3d8c3df
fix(skills): unify needs-human decision handoff ( #1492 )
2026-08-20 10:13:57 -07:00
9f55d68a96
fix(ce-resolve-pr-feedback): keep replies out of pending reviews ( #1491 )
2026-08-20 10:13:57 -07:00
Trevin Chow