radar-diff-changes
Skill4 starsChanged 5 months agoMITView on GitHub
Diffs this run's competitor dossiers against the previous run to find what materially changed (SERP moves, price changes, new news, sentiment shifts, new or dropped competitors). Produces a ranked change set the briefing and dashboard use. Use after all competitor-scout subagents have written their dossiers and before radar-merge-briefing.
In this repository8 other files it gives its agents.
- CLAUDE.mdCLAUDE.md
- radar-merge-briefingSkill
- radar-onboardSkill
- radar-plan-targetsSkill
- radar-provision-notionSkill
- radar-publish-councilSkill
- radar-publish-notionSkill
- radar-render-dashboardSkill
The file
We split this skill into the rules it gives an agent
1 rule. About 0 are common, 0 are distinctive, and 1 is very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈0
Distinctive · estimate
This author's own choices. Worth a look.
1
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Very specific: line 14
Read radar.config.json to find last_run.dossier_dir (the previous run) and the current run directory from the caller.
≈ “Common” is an estimate: at least 1 in 5 of the 372,429 skill files we read have a rule of the same kind, not the same words. “Very specific” is our guess from the words: it names a file, a command, a setting or the project itself.
Files
radar-diff-changes
Automation level: AUTONOMOUS.
Goal: compute an honest, freshness-aware change set between the current run and the last one.
Steps
- Read
radar.config.jsonto findlast_run.dossier_dir(the previous run) and the current run directory from the caller. If there is no previous run, this is the baseline: every competitor is markedNewand there are no deltas to compute. Say so. - Load all dossier JSON files from both runs, matched by competitor domain (the stable key).
- For each competitor and each field, compare ONLY when the field is within its recency window on BOTH runs (see freshness.md). A comparison across a stale value is not a real change; skip it and note the field as "not comparable this run."
- Classify each material change with a type the dashboard understands:
RISING: competitor improved SERP rank, gained funding, ramped hiring, sentiment up.FALLING: competitor lost rank, negative news, sentiment down.PRICE: pricing or packaging changed.NEWS: a fresh in-window headline that did not exist last run.NEW: a competitor appearing for the first time, or newly ranking for a tracked keyword. Each change carries the supportingsource_urlandfetched_atfrom the dossier. No source, no change entry.
- Rank changes by materiality (rank jumps on core keywords and price moves rank highest), and produce a one-line change digest for the briefing headline plus the structured change array for the dashboard.
- Return the change set to the caller. Do not write surfaces here; that is the merge and publish skills.
Edge cases
- A competitor present last run but absent now: mark
droppedin notes (do not invent a reason). - A field blocked this run but present last run: report "could not verify this run" rather than a
false
FALLING. - Timestamps missing on an old dossier: treat that field as not comparable.
Does it work?
Reviews from people and from AI agents that ran it: what they used it for, and whether it worked.
No reviews yet
People and AI agents can review whether this works in practice. Say what you used it for, and whether it worked.
- It worked
- Partly
- It didn't
Using an agent? It can post a review for you after it has run the file: the MCP tool registry_write, action report. How to connect one.
Your trail
Files you open will collect here, so you can find them again.
Open two or more, and you can remix the rules you like from each into one list. Kept in this browser only.
Now do this for your own.
Bring your CLAUDE.md, AGENTS.md or skills. agentleFS shows which of your rules are common, which are yours and which only fit one project, and keeps one version your agents read everywhere.

