Codex SPECTRE Implementation

Trigger: codex, spectre, codex install, sessionstart, agents.override, registry, spectre-learn, spectre-recall, hooks.json, config.toml, doctor Confidence: high Created: 2026-03-30 Updated: 2026-07-19 Version: 5

Current Design

Codex SPECTRE installs the workflow as Codex-native skills, subagent TOML configs, and generated SessionStart hooks. Project knowledge and session continuity are written into managed AGENTS.override.md blocks so hook output stays short.

Reusable project knowledge is both configured as normal skills and injected as a compact trigger registry:

• spectre-learn writes project skills under .agents/skills/{category}-{slug}/SKILL.md.
• The recall registry lives at .agents/skills/spectre-recall/references/registry.toon.
• spectre-recall is generated as an explicit search/load skill.
• Project installs sync .agents/skills/*/SKILL.md into Codex [[skills.config]].
• spectre-apply contains a {{REGISTRY}} placeholder that both the hook and src/lib/knowledge.js substitute before writing the managed knowledge block.
• bootstrap → handoff-resume → load-knowledge is the required SessionStart order.

Workflow task execution now uses a two-artifact contract:

• spectre-create_tasks writes {OUT_DIR}/specs/execute.md plus {OUT_DIR}/specs/tasks.json.
• execute.md is the compact primary-agent index (document manifest, task detail source, execution summary, wave plan, parent-task index, slicing rules).
• tasks.json is the full mutable detail/status source (meta + phases[]); primary execution/review/validation consumers should slice it by parent task id instead of reading the whole file.
• Do not reintroduce the old specs/tasks.md task-list flow or a Markdown fallback/converter.

Review gates use one cross-runtime contract:

• spectre-plan_review prefers the opposing CLI with an explicit high-effort model: Codex launches Claude Code with --model opus --effort high; Claude Code launches Codex with -m gpt-5.6-sol -c 'model_reasoning_effort="high"'. The launcher allows each attempt up to 20 minutes, but passes no duration guidance to the reviewer.
• spectre-task_review uses a focused medium-effort opposing-runtime contract: Codex launches Claude Code using --model opus --effort medium; Claude Code launches Codex using -m gpt-5.6-sol -c 'model_reasoning_effort="medium"'. The primary agent runs consumer-safety preflight, explicitly launches and monitors the reviewer for up to 20 minutes, validates the report, allows one repair, owns native fallback and write-back, and passes no duration guidance to the reviewer.
• spectre-code_review uses a high-effort opposing-runtime contract: Codex launches Claude Code using --model opus --effort high; Claude Code launches Codex using -m gpt-5.6-sol -c 'model_reasoning_effort="high"'.
• If the opposing runtime is unavailable or fails validation after one repair attempt, the gate dispatches one native reviewer with the same manifest, adversarial lenses, severity/evidence rules, exclusions, and report schema. This fallback does not block completion and must record its reason plus runtime/model metadata.
• spectre-code_review is an adversarial, evidence-gated review for correctness, regressions/integration, security, performance/reliability, overengineering, and test adequacy. It does not use subjective numeric scores.
• spectre-execute delegates its final cumulative review to spectre-code_review --orchestrated; do not reintroduce a separate final-review prompt inside execute.
• Canonical workflow skills live under plugins/spectre/skills/; regenerate plugins/spectre-codex/ and keep regression assertions in scripts/test_sync-codex.cjs aligned with these invariants.

Install Flow

1. src/main.js parses install codex, resolves scope, and switches CODEX_HOME to ./.codex for project installs.
2. installCodex() in src/lib/install.js copies generated Codex assets from plugins/spectre-codex/:
   • workflow skills into CODEX_HOME/skills/
   • agent TOML configs into CODEX_HOME/spectre/agents/
   • generated hooks into CODEX_HOME/spectre/hooks/
   • runtime helper scripts into CODEX_HOME/spectre/tools/
3. installCodex() removes the fork-era sibling runtime and agent tables before writing current [agents.spectre_*] definitions.
4. ensureSpectreHooksConfigured() enables features.hooks, features.skills, and features.multi_agent, then materializes generated SessionStart commands into CODEX_HOME/hooks.json without clobbering unrelated handlers.
5. For project installs, installProjectFiles() creates .spectre/manifest.json, initializes recall files, clears stale managed blocks, and calls syncProjectSkillsConfigured().
6. On SessionStart, the hooks refresh session and knowledge blocks only when the workspace has the relevant Spectre surface.

Key Files

• plugins/spectre/skills/ Canonical Claude/Codex-compatible workflow skill sources.
• plugins/spectre-codex/ Generated Codex bundle. Regenerate with npm run sync-codex -- --quiet.
• src/lib/install.js Main installer/uninstaller.
• src/lib/config.js Owns config.toml, hooks.json, agent tables, and project skill sync.
• src/lib/project.js Owns .spectre/manifest.json, handoff lookup, managed override blocks, and legacy cleanup.
• src/lib/knowledge.js Owns recall generation and Codex-side {{REGISTRY}} substitution.
• src/lib/doctor.js Verifies installed runtime/config state and reports stale hook remnants.

Common Tasks

Add or change a workflow skill

1. Edit plugins/spectre/skills/spectre-*/SKILL.md.
2. Run:

   npm run sync-codex -- --quiet
   npm run sync-codex -- --check --quiet
   

3. Run focused tests when installer or translator behavior changed:

   node --test src/install.test.js src/config.test.js scripts/test_sync-codex.cjs
   

4. If changing spectre-create_tasks task artifacts, also validate both task fixtures parse:

   node -e "JSON.parse(require('fs').readFileSync('plugins/spectre/skills/spectre-create_tasks/references/tasks.example.json','utf8'))"
   

Add a learned project skill and make sure Codex sees it

1. Write the skill under .agents/skills/{category}-{slug}/SKILL.md.
2. Register it in .agents/skills/spectre-recall/references/registry.toon.
3. Regenerate .agents/skills/spectre-recall/SKILL.md.
4. Refresh project install state:

   npx @codename_inc/spectre update codex --scope project --project-dir "$PWD"
   

Debug why a project skill is not being used

Check, in order:

1. The skill exists at .agents/skills/{name}/SKILL.md.
2. Its frontmatter description contains concrete trigger language.
3. config.toml contains a [[skills.config]] entry for the skill path.
4. If explicit search is needed, the registry entry exists in .agents/skills/spectre-recall/references/registry.toon.
5. hooks.json contains SessionStart commands for bootstrap.mjs, handoff-resume.mjs, and load-knowledge.mjs.
6. AGENTS.override.md contains an inlined registry and no raw {{REGISTRY}}.
7. Run:

   npx @codename_inc/spectre doctor codex --scope project
   

Expected Install Artifacts

After npx @codename_inc/spectre install codex --scope project, expect files like:

.codex/config.toml
.codex/hooks.json
.codex/skills/spectre-apply/SKILL.md
.codex/skills/spectre-scope/SKILL.md
.codex/spectre/hooks/hooks.json
.codex/spectre/hooks/scripts/bootstrap.mjs
.codex/spectre/hooks/scripts/handoff-resume.mjs
.codex/spectre/hooks/scripts/load-knowledge.mjs
.codex/spectre/hooks/scripts/register_learning.mjs
.codex/spectre/agents/dev.toml
.spectre/manifest.json
.agents/skills/spectre-recall/SKILL.md
.agents/skills/spectre-recall/references/registry.toon

Do not reintroduce:

• spectre-guide
• spectre-evaluate
• spectre-architecture_review
• fork-name CLI aliases or duplicate runtime trees
• raw {{REGISTRY}} in generated or installed context
• startup payloads in additionalContext; use managed AGENTS.override.md blocks