record-observation
Skill0 starsChanged yesterdayMITView on GitHub
Use this skill to record an undocumented bug, environment quirk, runtime edge-case, or non-obvious solution into the repository's `.agents/project_knowledge/observations.md` table.
In this repository19 other files it gives its agents.
- AGENTS.mdAGENTS.md
- templates/AGENTS.mdAGENTS.md
- decommission-meta-projectSkill
- ground-meta-projectSkill
- init-meta-projectSkill
- manage-meta-projectSkill
- scan-meta-projectSkill
- sync-architectureSkill
- tune-meta-projectSkill
- update-meta-projectSkill
- decommission-meta-projectSkill
- ground-meta-projectSkill
- init-meta-projectSkill
- manage-meta-projectSkill
- record-observationSkill
- scan-meta-projectSkill
- sync-architectureSkill
- tune-meta-projectSkill
- update-meta-projectSkill
The file
We split this skill into the rules it gives an agent
3 rules. About 0 are common, 1 is distinctive, and 2 are very specific to this project.
≈0
Common · estimate
Many files like this have a rule of this kind.
≈1
Distinctive · estimate
This author's own choices. Worth a look.
2
Very specific
Tied to their project's names, files or tools. Probably skip.
- Common
- Distinctive
- Very specific
- Very specific: line 25
Read the recent entries of .agents/project_knowledge/observations.md.
- Very specific: line 31
No canonical copy → schedule migration (move legacy ➔ scripts/).
- Distinctive: line 55
If the number of rows in the table exceeds 20–30, locate the compaction utility deterministically:
≈ “Common” is an estimate: at least 1 in 5 of the 364,980 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
Record Observation in Repository Knowledge Base
This skill ensures disciplined, standardized addition of entries to the project observation registry without corrupting document structure or causing context bloat.
When to invoke skill
- The agent or user encounters an unexpected environment, build, or runtime error that was successfully resolved.
- A critical behavioral nuance of a third-party library or internal API is discovered that is not documented in official project documentation.
- The user instructs: "record observation", "add to knowledge base", "log gotcha", "record finding".
Execution steps
- Check for duplicates:
Read the recent entries of.agents/project_knowledge/observations.md. Ensure that a similar observation has not already been logged. If an entry already exists, do not create duplicate rows. Execute standardized CLI utility (
record_obs.py):
Locate the runtime utility deterministically:- Canonical location first:
scripts/record_obs.py. - Legacy
.agents/scripts/record_obs.pyis supported ONLY as an explicit migration candidate: if found, compare with the canonical copy.- No canonical copy → schedule migration (move legacy ➔
scripts/). - Both exist and are IDENTICAL → migrate (delete legacy) at next update.
- Both exist and DIVERGE → HALT and report both paths with hashes; never silently use the first found.
- No canonical copy → schedule migration (move legacy ➔
Record the finding atomically with the resolved copy:
python scripts/record_obs.py --category "<category>" --path "<path_or_component>" --issue "<issue_summary>" --solution "<workaround_or_solution>" --actor-id "<actor_id>"- Parameters:
--category: One of the canonical categories:boundaries,ci,decommission,incident,orchestration,runtime,git,init,testing,architecture.--path: Component, script, or configuration path.--issue: Brief summary of the bug or quirk (pipes|and newlines are safely escaped automatically).--solution: Practical workaround or fix.--actor-id: Optional qualified actor identifier (e.g.<project_id>::<harness_id>::<installation_id>::<role_id>) for action attribution (inserts[actor=<actor_id>]into the issue field).--pin: Optional flag for permanent invariants that must never be rotated bycompact_memory.py(adds[pin]suffix, e.g.[ci][pin]).--obs-file: Optional custom path (defaults to.agents/project_knowledge/observations.md).
- Canonical location first:
- Fallback manual entry (only if CLI script unavailable):
Format the observation strictly as a single Markdown table line with escaped pipes\|:| YYYY-MM-DD | [<category>] | <Path_or_component> | <Brief issue summary/fact> | <Solution or workaround> | Registry size control (Compaction Protocol):
If the number of rows in the table exceeds 20–30, locate the compaction utility deterministically:- Canonical location first:
scripts/compact_memory.py. - Legacy
.agents/scripts/compact_memory.pyis supported ONLY as an explicit migration candidate: if found, compare with the canonical copy.- No canonical copy → schedule migration (move legacy ➔
scripts/). - Both exist and are IDENTICAL → migrate (delete legacy) at next update.
- Both exist and DIVERGE → HALT and report both paths with hashes; never silently use the first found.
- No canonical copy → schedule migration (move legacy ➔
Invoke the automated compaction utility with the resolved copy:
python scripts/compact_memory.py --max-observations 20 --keep-observations 10This automatically rotates older records into
.agents/project_knowledge/archive/observations_YYYYMM.md, keeping the primary file lightweight and preventing context rot. (Ifcompact_memory.pyis not present inscripts/, manually archive older resolved records into.agents/project_knowledge/archive/observations_YYYYMM.md).- Canonical location first:
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.

