request-curation
Skill8 starsChanged 22 days agoMITView on GitHub
Use when the user wants to refine what kinds of items matter within an existing topic, add or remove tracked angles, or turn digest feedback into durable brief rules. Trigger for Chinese feedback like "这条以后少推", "这类以后多推", "加上价格变化", or "别再推这种".
In this repository7 other files it gives its agents.
- AGENTS.mdAGENTS.md
- verify-skryaSkill
- deep-analysisSkill
- digestSkill
- skryaSkill
- source-curationSkill
- topic-curationSkill
The file
We split this skill into the rules it gives an agent
11 rules. About 0 are common, 8 are distinctive, and 3 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈8
Distinctive · estimate
This author's own choices. Worth a look.
3
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Very specific: line 11
Always resolve an explicit internal topic-id before reading or writing files, but do not ask nontechnical users for raw ids when a natural …
- Very specific: line 12
Before reading or writing topic files, resolve the Skrya data root.
- Distinctive: line 19
If the user is reacting to ranking or exclusions in a recent digest, also read:
- Distinctive: line 46
check latest digest for whether X was present under another wording
- Distinctive: line 47
check latest ingest artifact for whether X was collected but ranked out
- Distinctive: line 48
check source coverage and runtime retrieval query terms for whether X could be collected
- Distinctive: line 49
check ranking and exclusion rules for whether X was filtered or down-ranked
- Very specific: line 59
Only touch digest.md if the requested change is really about ranking or exclusion policy rather than tracked angles.
- Distinctive: line 63
Keep the conversation natural and short.
- Distinctive: line 64
Do not expose internal debug fields or config jargon unless needed.
- Distinctive: line 70
After this skill completes:
≈ “Common” is an estimate: at least 1 in 5 of the 350,495 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
<!-- AUTO-GENERATED from request-curation/SKILL.md.tmpl; regenerate with `python -m skrya_orchestrator.main build-skill-pack --root . --host all` -->Request Curation
Translate user feedback about what should matter in a configured topic into durable request language.
Always resolve an explicit internal topic-id before reading or writing files, but do not ask nontechnical users for raw ids when a natural topic name can be mapped.
Before reading or writing topic files, resolve the Skrya data root. Topic configuration and generated artifacts live under <skrya-data-root>/, not necessarily under the skill repository.
Read First
<skrya-data-root>/topics/<topic-id>/topic.json<skrya-data-root>/topics/<topic-id>/brief.json
If the user is reacting to ranking or exclusions in a recent digest, also read:
<skrya-data-root>/topics/<topic-id>/digest.md<skrya-data-root>/runs/<topic-id>/latest-digest.mdwhen available<skrya-data-root>/runs/<topic-id>/latest-digest-events.jsonwhen available<skrya-data-root>/runs/<topic-id>/ingest/latest-ingest.jsonwhen available
Also inherit any applicable workspace defaults from AGENTS.md, CLAUDE.md, or the equivalent repository instruction file.
Use This Skill When
- the user wants to add a new tracked angle
- the user wants to narrow topic scope
- the user wants to exclude a class of items from future digests
- the user wants to rebalance what is ranked higher or lower
- the user points at one digest item as an example of what should or should not be surfaced again
- the user says "这个很重要"
- the user asks "为什么没有 ..." or reports a missed item
Core Rule
Convert one-off user feedback into durable brief.json request language instead of copying the example literally.
Missed Item Feedback
For "为什么没有 X", treat it as 漏报诊断 before editing anything:
- check latest digest for whether X was present under another wording
- check latest ingest artifact for whether X was collected but ranked out
- check source coverage and runtime retrieval query terms for whether X could be collected
- check ranking and exclusion rules for whether X was filtered or down-ranked
- propose the smallest durable change to
brief.json,digest.md, or sources after explaining the likely failure point
Required Behavior
- Interpret the user request as a persistent preference, not a one-off chat reply.
- Draft the smallest durable wording that captures that preference.
- Show the proposed wording to the user in concise natural Chinese.
- Confirm before writing any file.
- Prefer updating
brief.json. - Only touch
digest.mdif the requested change is really about ranking or exclusion policy rather than tracked angles.
Output Rules
- Keep the conversation natural and short.
- Do not expose internal debug fields or config jargon unless needed.
- File names are internal execution details; do not show file names in normal user-facing replies unless the user asks for implementation details.
- Preserve the user's intent, but normalize the language into stable configuration text.
Success Criteria
After this skill completes:
brief.jsonreflects a durable tracking request- one-off examples have been generalized into stable rules
- the user has explicitly confirmed the wording before it is written
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.

