ariadne:knowledge-capture
SkillDocs & knowledgeUse when links, documents, PDFs, transcripts, screenshots, meeting notes, observations, brain dumps, or other material must be compiled into an explicitly selected Markdown knowledge-vault destination.
Available today. Use it from your connected AI after setup.
No other account needed.
Connect ahel once, and every AI you use reads what you have installed.
Then ask your AI: use the ariadne:knowledge-capture skill
What this skill tells your AI
The instructions your AI receives, as published by pariyar07/ariadne in skills/knowledge-capture/SKILL.md and read by ahel’s review.
Compile material after its destination and write set are known. This skill does not select a research scope, create research topology, classify evidence independently, or judge synthesis disposition.
Authorization
- Require a known destination and explicit
allowed_write_setbefore writing. In a multi-scope vault, the target must be named or confirmed in the current turn. - Read root instructions and every ancestor scope instruction through the target.
- Read relevant indexes, agent navigation, routing, local ingest workflow, and folder hubs progressively.
- Keep every mutation inside the write set. Parent/root navigation, sibling scopes, and downstream destinations require individually named paths.
For research, consume the full ariadne:research-ingest handoff envelope without changing it:
target_scope
research_boundary
allowed_write_set
source_type
evidence_role
derived_from
inquiry_links
pipeline_state
requested_operation
If a research envelope is missing, ambiguous, or requests paths outside the set, return to ariadne:research-ingest. Route missing topology to ariadne:research-pipeline rather than creating it silently.
Capture And Compile
- Capture one canonical raw artifact when the source must remain inspectable. For research, preserve
source_type,evidence_role,derived_from, andresearch_boundaryexactly from the handoff. Map handoffinquiry_linksto artifact fieldinquiries; use an empty flat list when no inquiry is known. - Separate observed or source claims from interpretation.
- Compile the smallest durable note or notes supported by the material.
- For compiled research, map handoff
inquiry_linksto canonical frontmatter fieldinquirieswithout inventing an inquiry, and link artifacts downstream to upstream withderived_from. - Link artifacts from authorized local hubs. Report any out-of-set navigation obligation instead of expanding scope.
- Add follow-up items only inside the allowed write set.
- If compiled research may affect current understanding, hand it to
ariadne:research-synthesis. This skill does not judge synthesis disposition.
Generated analysis and derivative copies retain upstream provenance and do not count as independent corroboration. Capture cross-scope material once; do not duplicate raw evidence.
Minimum Research Raw Schema
Write research raw artifacts with at least this flat schema, plus local title, origin, locator, and date fields required by ancestor instructions:
type: raw-source
research_boundary: "[[Domains/Example/Research/00 Example Research Boundary]]"
source_type: meeting
evidence_role: first-party-evidence
origin: Product interview
locator: not-applicable
captured: 2026-07-15
compilation_status: pending
derived_from: []
inquiries: []
Allowed compilation states are pending, compiled, source-only, or needs-review. Use compiled only when at least one derived note links back to the raw artifact. Do not change the handoff classification merely to satisfy a local template.
Promotion Writes
Accept a promotion candidate from ariadne:research-synthesis only when its destination is named or confirmed in the current turn and included in the write set. Create or update the destination note with research_basis or equivalent derived_from links. Do not move, duplicate, or silently rewrite the canonical research.
Validation
Before finishing, confirm that every write was authorized, raw material is compiled or intentionally source-only, provenance is present, claims and interpretation remain distinct, important artifacts are locally discoverable, and focused validation passes or its warnings are reported.
Use ariadne:research-ingest for research target resolution and classification, ariadne:research-pipeline for topology, ariadne:research-synthesis for dispositions, ariadne:navigation for separately authorized structural routes, and ariadne:maintenance for broad vault health.
Signals
- GitHub stars
- 20
- Forks
- 2
- Last commit
- Aug 2026
Advanced
- Catalog kind
- skill
- Gateway key
ariadne-knowledge-capture- Source
- github.com/pariyar07/ariadne