Regulations.gov Comments Fetch

Core Goal

• Fetch comments data from https://api.regulations.gov/v4/comments.
• Filter by configurable time windows (lastModifiedDate or postedDate).
• Apply optional filters (agencyId, commentOnId, searchTerm).
• Return machine-readable JSON and optionally save JSONL records locally.
• Keep execution observable with structured logs and optional log file.

Required Environment

• Configure runtime with environment variables (see references/env.md).
• Start from assets/config.example.env.
• Load env values before running commands:

set -a
source assets/config.example.env
set +a

Workflow

1. Validate effective configuration.

python3 scripts/regulationsgov_comments_fetch.py check-config --pretty

2. Dry-run to verify query parameters without making remote calls.

python3 scripts/regulationsgov_comments_fetch.py fetch \
  --filter-mode last-modified \
  --start-datetime 2026-03-10T00:00:00Z \
  --end-datetime 2026-03-10T23:59:59Z \
  --max-pages 2 \
  --dry-run \
  --pretty

3. Fetch comments with structure validation and logs.

python3 scripts/regulationsgov_comments_fetch.py fetch \
  --filter-mode last-modified \
  --start-datetime 2026-03-10T00:00:00Z \
  --end-datetime 2026-03-10T23:59:59Z \
  --agency-id EPA \
  --max-pages 3 \
  --max-records 300 \
  --output-dir ./data/regulationsgov-comments \
  --quarantine-dir ./data/regulationsgov-comments-quarantine \
  --log-level INFO \
  --log-file ./logs/regulationsgov-comments-fetch.log \
  --pretty

Built-in Robustness

• Retry transient failures (429/500/502/503/504) with exponential backoff.
• Respect Retry-After on rate-limit responses and stop if it exceeds configured cap.
• Throttle request frequency with a minimum interval between requests.
• Enforce safety caps:
  • --max-pages limited by REGGOV_MAX_PAGES_PER_RUN
  • --max-records limited by REGGOV_MAX_RECORDS_PER_RUN
• Validate transport and payload:
  • HTTP/content-type checks
  • UTF-8 + JSON parse checks
  • JSON:API-style data/meta structure checks
  • comment item field checks (id, type, datetime field formats)
  • optional issue quarantine (--quarantine-dir)

Scope Decision

• Keep one atomic endpoint implementation: comments list fetch.
• Keep operations request-driven; do not add built-in scheduler/polling loops.
• Preserve simple downstream integration by returning JSON and optional JSONL artifact.

References

• references/env.md
• references/regulationsgov-api-notes.md
• references/regulationsgov-limitations.md
• references/openclaw-chaining-templates.md

Script

• scripts/regulationsgov_comments_fetch.py

OpenClaw Invocation Compatibility

• Keep trigger metadata in name, description, and agents/openai.yaml.
• Invoke in prompts with $regulationsgov-comments-fetch.
• Keep the skill atomic: each invocation handles one configured window.
• Use script parameters for fetch conditions (--filter-mode, time window, page caps).
• If periodic polling is needed, let OpenClaw orchestrate repeated invocations externally.

OpenClaw Prompt Templates

Use these templates directly in OpenClaw and only replace bracketed placeholders.

1. Recon (query plan check)

Use $regulationsgov-comments-fetch.
Run:
python3 scripts/regulationsgov_comments_fetch.py fetch \
  --filter-mode last-modified \
  --start-datetime [YYYY-MM-DDTHH:MM:SSZ] \
  --end-datetime [YYYY-MM-DDTHH:MM:SSZ] \
  --max-pages [N] \
  --dry-run \
  --pretty
Return only the JSON result.

2. Fetch (time-window comments)

Use $regulationsgov-comments-fetch.
Run:
python3 scripts/regulationsgov_comments_fetch.py fetch \
  --filter-mode last-modified \
  --start-datetime [YYYY-MM-DDTHH:MM:SSZ] \
  --end-datetime [YYYY-MM-DDTHH:MM:SSZ] \
  --agency-id [OPTIONAL_AGENCY] \
  --max-pages [N] \
  --max-records [M] \
  --output-dir [OUTPUT_DIR] \
  --quarantine-dir [QUARANTINE_DIR] \
  --pretty
Return only the JSON result.

3. Validate (quality gate)

Use $regulationsgov-comments-fetch.
Run:
python3 scripts/regulationsgov_comments_fetch.py fetch \
  --filter-mode posted \
  --start-date [YYYY-MM-DD] \
  --end-date [YYYY-MM-DD] \
  --max-pages 1 \
  --max-records 50 \
  --pretty
Check validation_summary.total_issue_count and stop_reason.
Return JSON plus one-line pass/fail verdict.