Skip to content

hashgraph-online/awesome-codex-plugins / plugins/boshu2/agentops/skills/skill-builder

skill-builder

Verified

Create, repair, audit or consolidate agent skills (SKILL.md packages). Use when: writing or fixing a skill, its description or structure. Not for one-off lessons; use Memory.

references/codex-parity.md

View on GitHub
# Codex parity

Codex loads the canonical `skills/<name>/` package directly. The Codex plugin
manifest points at `./skills`; a source checkout can instead link the same
directories with `ao skills link --dest ~/.codex/skills`. The bare
`ao skills link` links into every detected runtime, `~/.claude` and `~/.agents`
included. Use the link instead of the plugin, never alongside it: both together
list every skill twice. There is no generated Codex copy, no override layer and
no `prompt.md`, so one source package has to read correctly on every host.

## What Codex reads

- `SKILL.md`. The `name` and `description` drive discovery. Codex ignores the
  AgentOps host fields. It refuses a skill whose frontmatter repeats a key, has
  no `description`, or has a `name` longer than 64 characters.
- `agents/openai.yaml`, when present: display metadata, tool dependencies and
  the invocation policy. Codex does not read `disable-model-invocation` from
  `SKILL.md`.
- Every `SKILL.md` below `skills/`. A nested one is loaded as a skill of its
  own, so fixtures and scaffolds live outside the tree.

## Explicit-only skills

A skill marked `disable-model-invocation: true` needs the matching Codex policy
in its own `agents/openai.yaml`:

```yaml
policy:
  allow_implicit_invocation: false
```

Nothing derives this file. Without it, or when it does not parse, Codex selects
the skill implicitly.

## One body for every host

- Refer to another skill by name or relative link, not by a host's invocation
  syntax (`/name`, `$name` or a `Skill(...)` call).
- Keep host-only tool names and installed-skill paths (`~/.claude/...`,
  `~/.codex/...`) out of the shared flow. When a step differs by host, say which
  host the sentence is for.
- Name a bundled file by a literal `<skill-dir>` that SKILL.md defines once
  before its first command, and run scripts through their interpreter,
  for example `bash "<skill-dir>/scripts/x.sh"`. Not `${CLAUDE_SKILL_DIR}`:
  Codex and reference files leave it unexpanded, so the command runs
  `/scripts/x.sh`.
- A skill that documents several runtimes names each one plainly.

## Check

AgentOps source checkout only, from the repository root:

```bash
bash scripts/validate-codex-api-conformance.sh
```

It checks the loader facts above and the explicit-only policy. It does not
prove that Codex selected or followed the skill; that needs a session on the
host.
rea-analysis-optimizationmorluto/reaOptimize REA providers and shared analysis workflows for agent context, latency, CPU, and memory while preserving correctness and Evidence. Use for analysis bottlenecks and cache or query design; CI/test-suite optimization belongs to rea-ci-tests-audit.Stars: 469kUpdated Claude CodeCodexVerifiedInstall
rea-ci-tests-auditmorluto/reaAudit REA CI and test suites for correctness, fixture drift, false passes, and measured performance; use for suite or workflow investigations and efficiency changes, rather than ordinary feature testing or a single test failure.Stars: 469kUpdated Claude CodeCodexVerifiedInstall
rea-tool-designmorluto/reaAudit, investigate, design, or change REA investigation tools, CLI/MCP contracts, provider capabilities, and Evidence semantics.Stars: 469kUpdated Claude CodeCodexVerifiedInstall
ponytail-auditDietrichGebert/ponytailQuality audit of the whole repo: bugs, security, real load, missing tests, speed, and what to delete. Most important first.Stars: 194kUpdated Claude CodeCodexVerifiedInstall