scrape-competitor-pricing-weekly

Backed by: agent-browser MCP for fetch + state/lib/log.ts for the eval row.

Cron job (Saturday 9 AM). Checks competitor pricing pages for changes.

Steps

1. Read the competitor list from sources/competitors.json (create if missing).
2. For each competitor URL, fetch the pricing page via agent-browser.
3. Extract pricing tiers and amounts.
4. Compare against last week's snapshot in state/log/competitor-pricing/.
5. If any prices changed, log the delta.
6. Save this week's snapshot.
7. append("scrape-competitor-pricing-weekly", { ok, checked, changed }) via state/lib/log.ts.

Eval

Score 1.0 if all competitors checked and snapshot saved. Score 0.5 if some competitors checked. Score 0.0 if no competitors could be reached.

Rubric

criteria:
  - name: all_competitors_processed
    kind: judge
    check: "The skill logs indicate that pricing pages for all competitors listed in 'sources/competitors.json' were fetched and processed."
  - name: new_snapshot_saved
    kind: deterministic
    check: "A new pricing snapshot for the current week is present in 'state/log/competitor-pricing/'."
  - name: pricing_changes_logged
    kind: judge
    check: "If any pricing changes were detected by comparing against last week's snapshot, a clear delta was logged, including which competitor and specific changes."
  - name: eval_row_appended
    kind: deterministic
    check: "The log via 'state/lib/log.ts' contains an entry for 'scrape-competitor-pricing-weekly' with 'ok', 'checked', and 'changed' fields."

scrape-competitor-pricing-weekly - loader

Per-turn rules for the scrape-competitor-pricing-weekly skill. Full reference: state/skills/scrape-competitor-pricing-weekly/SKILL.md.

Critical Rules

• ALWAYS read the competitor list from sources/competitors.json (human-owned). If missing, CREATE it - do not invent competitors inline.
• ALWAYS save this week's snapshot to state/log/competitor-pricing/ BEFORE comparing - partial run with no snapshot loses the baseline for next week.
• NEVER fan out parallel agent-browser fetches against one session - DOM-mutation agents drop writes silently. Serialize.
• Score 0.5 if some competitors fetched (partial). Score 0.0 only if NONE could be reached (likely auth or network).

Commands

| ui dashboard | state/skills/scrape-competitor-pricing-weekly/resources/ui.openui | |invoke: cron-driven; manual via state/lib/log.ts append("scrape-competitor-pricing-weekly", { ok, checked, changed }) |fetch: agent-browser MCP (the documented exception lives in browser libs, not snappy-os core) |eval log: state/log/evals.ndjson (skill: "scrape-competitor-pricing-weekly")

OpenUI Resource

• Skill-owned OpenUI Lang resource: state/skills/scrape-competitor-pricing-weekly/resources/ui.openui. Read it before rendering or editing this skill's generated component surface.
• Treat this resource as a first-class artifact of the skill, not a generic chat response. Improve it when the skill's user-facing output needs to become richer.
• System resources compose OpenUI primitives and inherit SnappyChat tokens. Use ui_contract: branded in SKILL.md only for deliberate platform or client visuals.

Known Pitfalls

• sources/competitors.json is human-owned (per program.md three-layer rule). Do not auto-edit; create the file if absent and tell the operator.
• Pricing pages move tier names around between weeks - extract by amount, not by tier label.
• Cron fires Saturday 9 AM. Manual triggers should also tag the eval row so the cron history stays clean.

Self-Test

An agent reading this should correctly:

1. [ ] Refuse to invent competitors; create sources/competitors.json instead
2. [ ] Save this week's snapshot before comparing to last week
3. [ ] Serialize browser fetches (no parallel)

Self-report

If this loader fell short, append a line:

echo "[$(date -u +%FT%TZ)] scrape-competitor-pricing-weekly: <what was missing>" >> state/log/loader-feedback.log

<!-- kernel-ok: historical attribution for the PID-loader pattern. -->

Self-correcting loader (PID feedback)

This is the rule that makes loaders learn. Ported from snappy-kernel, where it demonstrably worked: the snappy-dom-cartographer skill mapped Skool admin UI once, every later ship-agent inherited the full map for free.

Before you finish, do two things:

1. Fix gaps (P - proportional). MANDATORY. If this loader didn't cover your case - if you had to read another file, run an undocumented command, work around a wrong selector, or discover a quirk - you MUST attempt an Edit to this AGENTS.md before you log.

• Default action: edit this file. Surgical. Add the missing line, table

row, or code example where it belongs. No restructuring.

• Only LOGGED is allowed when: the fix needs >10 lines, spans multiple

files, or requires a structural rewrite. In that case the state/regen/drain.sh queue picks it up asynchronously.

• "I didn't have time" / "it's minor" / "the next agent will figure it out"

are NOT valid reasons. Minor gaps compound - that's the whole point of this loop.

• Concrete example of a valid P-fix: discover agent-browser screenshot

takes path positionally not as --output <path>. One line in a table. Edit it in. Do not just log it.

• The goal: the next agent never has to leave the loader.

2. Log the result.

echo "[$(date -u +%FT%TZ)] scrape-competitor-pricing-weekly: <what was missing or fixed> [FIXED|LOGGED]" >> state/log/loader-feedback.log

• FIXED = you patched this loader inline (P-fix).
• LOGGED = too large for inline; the PostToolUse enqueue + Stop-hook drain

will rewrite the loader from scratch on next session-end.

Do not skip this. Every agent run must leave the system better than it found it. The loader is the setpoint; you are the sensor; the gap is the error signal; closing the gap is the correction.