Skip to content

docs: fix drift in OpenAI Codex - #259

Draft
jack-arturo wants to merge 1 commit into
mainfrom
docs/audit-platforms-codex-20260716
Draft

docs: fix drift in OpenAI Codex#259
jack-arturo wants to merge 1 commit into
mainfrom
docs/audit-platforms-codex-20260716

Conversation

@jack-arturo

Copy link
Copy Markdown
Member

Routine docs-accuracy audit of src/content/docs/docs/platforms/codex.md against verygoodplugins/mcp-automem@main (never previously reviewed).

Claim Current state Fix
"The codex CLI command generates an AGENTS.md section with placeholders: {{PROJECT_NAME}} ... {{CURRENT_MONTH}} → current month in YYYY-MM format" src/cli/codex.ts only builds vars = { PROJECT_NAME: projectName } — no CURRENT_MONTH var is ever passed to replaceTemplateVars. A test elsewhere (tests/cli/smoke.test.ts) explicitly asserts a sibling template does not contain {{CURRENT_MONTH}}. This is the same fabricated-placeholder bug already found and fixed on best-practices/progressive-disclosure.md (PR #232). Removed the {{CURRENT_MONTH}} bullet; noted explicitly that only {{PROJECT_NAME}} is substituted.

Everything else checked out: the CLI Command Reference table (--name, --dry-run, --rules <path>) matches parseCommonFlags/codex.ts exactly, and the Content Size Guidelines table (150–300 target, 500-char auto-summarize threshold, 2000-char hard limit) matches automem/config.py's MEMORY_CONTENT_SOFT_LIMIT/MEMORY_CONTENT_HARD_LIMIT defaults.

Open question (not fixed, flagging per audit guidelines): the actual generated templates/codex/memory-rules.md (v0.15.0) is a much more detailed ruleset than what "Memory Operation Patterns" describes — e.g. it does two-phase session-start recall (separate preference vs. task-context calls with fixed limits 20/30 and a 90-day window) rather than the single combined recall_memory call shown in the doc, plus an "atomic ritual" and a mandatory-association table. The doc's simplified version uses valid, real tool parameters throughout, so it isn't concretely wrong — but it may be describing an older template generation or a deliberately simplified explanation rather than the literal current template. Didn't rewrite this since it would be a large structural change beyond a targeted drift fix; flagging for a maintainer call on whether the page should be brought in line with the current template text.

Verified against: verygoodplugins/mcp-automem@946f9e5 (src/cli/codex.ts, src/cli/host-toolkit.ts, templates/codex/memory-rules.md, tests/cli/smoke.test.ts)


Generated by Claude Code as part of the recurring automem-docs-review routine.


Generated by Claude Code

Routine docs-accuracy audit against verygoodplugins/mcp-automem@main.

Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01FX7QQ4GSzzAbonMyeNDYzJ
@cloudflare-workers-and-pages

Copy link
Copy Markdown

Deploying automem-website with  Cloudflare Pages  Cloudflare Pages

Latest commit: 8ec7762
Status: ✅  Deploy successful!
Preview URL: https://abd0022a.automem-website.pages.dev
Branch Preview URL: https://docs-audit-platforms-codex-2.automem-website.pages.dev

View logs

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment

Labels

None yet

Projects

None yet

Development

Successfully merging this pull request may close these issues.

2 participants