2655 Contextual AI Documentation Optimizer Adapter 394973bd

contextual-ai-documentation-optimizer Adapter

Wrapped Component

plugins/plugin-creator/agents/contextual-ai-documentation-optimizer.md

This agent produces a multi-section output: RT-ICA assessment, optimized content, CoVe verification, and token impact report. Output is a STATUS: DONE or BLOCKED block with the optimized content inline or written to file.

Native Output Contract

The agent produces:

1. RT-ICA pre-check assessment
2. Optimized file content (written in-place or to specified path)
3. CoVe post-check verification (did the output meet the objectives?)
4. Token impact report (before/after token counts)
5. STATUS: DONE or BLOCKED

Adapter Transformation

The adapter maps the native output to optimization-block-v1:

STATUS: DONE|BLOCKED|FAILED
SUMMARY: [what was optimized — extracted from agent's STATUS summary]
ARTIFACTS:
  - {optimized-file-path}
VALIDATION:
  - frontmatter-validator: PASS|FAIL
  - prompt-structure-validator: PASS|FAIL
DIFF:
  tokens_before: N
  tokens_after: N
  delta: +N / -N
  prohibitions_converted: N
NOTES: [CoVe verification notes, if any issues found]

Validators Added by Adapter

• frontmatter-validator — re-runs after optimization to confirm frontmatter is still valid
• prompt-structure-validator — checks that prohibition patterns were converted to positive alternatives

BLOCKED Mapping

If the agent returns BLOCKED (typically due to RT-ICA pre-check failure), the adapter preserves the BLOCKED status. The RT-ICA failure reason is placed in NOTES. The user must provide the missing input before re-running.

Token Impact Data

The adapter extracts token counts from the agent's token impact report. If the optimized content increased token count by more than 10%, the NOTES field includes a flag: "Token count increased — review before committing."