- Home
- Skills
- Agents & Orchestration
- Second-Brain Knowledge Base Orchestrator
More screenshots
Works with the AI Agents you already use
Second-Brain Knowledge Base Orchestrator
Maintain durable, evidence-backed project knowledge across agent sessions with structured capture and reconciliation.
$3
Second-Brain Knowledge Base Orchestrator
Example session with this skill installed
Use Knowledge Base Orchestrator to ingest docs/api-v2-draft.md into this project’s existing knowledge base.
Keep this task focused on authentication: token rotation and lifetime, the OAuth2 refresh grant, and the proposed /revoke endpoint.
- Read the project’s KB instructions and index first. Reuse existing locations and conventions.
- Preserve the source and record its provenance, revision date and full content hash.
- Compare the draft with retained requirements and approved decisions, including
kb/decisions/auth-004.md. Distinguish proposals from approved behavior; check whether rotation frequency and absolute lifetime actually conflict. - Update affected knowledge pages, the index and the ingestion log. Retain dated history and flag missing requirements without inventing defaults.
- Confirm the active code candidate before inspecting authentication code. Keep implementation observations separate from policy requirements. Do not change application code.
Return a concise update report covering sources, pages changed, supported findings, conflicts, gaps, checks actually run and the next action. Clearly distinguish prepared knowledge from verified retrieval or implementation.
- Read your context and instructions
- Compiled the second-brain knowledge base
Project Knowledge Updated
I integrated the API v2 draft into the project’s retained knowledge. Future sessions can now find the proposed authentication changes, their relationship to existing decisions, and the remaining questions through the project index.
Illustrative example output.
Knowledge retained for future sessions
| Topic | Retained knowledge |
|---|---|
| Existing authentication decision | The approved decision specifies a 24-hour token limit. Its original source and rationale remain linked. |
| Proposed change | The API v2 draft introduces 15-minute token rotation, a refresh grant and a /revoke endpoint. These are recorded as proposals. |
| Connection between sources | Token rotation frequency and maximum token lifetime may be separate requirements. The knowledge page explains this distinction instead of treating the two values as an established contradiction. |
| Open question | The draft does not define revocation rate limits. This remains explicitly unknown. |
Where that knowledge lives
- Preserved source: the original API draft remains unchanged and traceable through the source register.
- Authentication overview: connects the existing decision, proposed flow and unresolved questions.
- Token concept page: explains rotation, lifetime and how the two source documents relate.
- Project index: links to these pages so another session can discover them.
- Knowledge log: records what was added, which pages changed and why.
Consistency update: Related pages now distinguish approved behavior from proposed changes. The unresolved question is recorded once and linked wherever relevant.
The draft has been integrated into related knowledge rather than left as an isolated summary. The original decision remains available alongside the proposed changes.
What a future session can answer
Question: “What do we know about the proposed authentication changes?”
Retained answer: “The draft proposes 15-minute token rotation, refresh grants and revocation. The existing approved decision specifies a 24-hour token limit. Whether these coexist needs clarification; revocation rate limits are unspecified.”
That answer can be located through the index, with links back to the supporting sources, without reconstructing the discussion from previous chats.
Next useful action
Clarify how rotation and lifetime should interact. When that answer arrives, preserve it as evidence, update the connected pages and record the resolved question. Useful knowledge accumulates in the project instead of disappearing into another conversation.
This knowledge update does not establish that the application implements the proposed behavior.
Connects securely to your tools. The creator never sees your data.
About this skill
Stop losing critical project context between agent sessions. This skill manages a durable project knowledge base (KB) that agents use to find, verify, and update information without drifting from the original objective. It solves the problem of "agent amnesia" by establishing a formal system for capturing evidence and reconciling conflicting claims.
What it does
- Source preservation records origin, dates, and hashes for immutable evidence snapshots to prevent instruction injection.
- Knowledge compilation updates concept, entity, and decision pages by integrating new claims rather than piling up summaries.
- Evidence-backed querying retrieves specific source spans and highlights gaps when compiled knowledge is insufficient for an answer.
- Conflict reconciliation compares new claims against existing data using authority and effective dates to manage contradictions.
- Project state tracking maintains an index of milestones, verified results, and open questions to ensure session continuity.
How it works
- Establish the workspace by reading the schema, index, and current objectives to confirm the active repository branch.
- Capture new evidence by preserving the source and linking claims to specific supporting spans in the knowledge pages.
- Maintain health by linting touched pages for stale statements, broken links, or unsupported claims.
- Report progress by summarizing added knowledge, identified conflicts, and the next required action.
Frameworks & tools
This skill is framework-agnostic and works with any text-based repository structure. It utilizes standard Markdown for documentation and relies on Git or local file systems for versioning and source hashing.
Why this beats prompting it yourself
Generic prompts often lead to agents hallucinating "memory" or overwriting old requirements with new, less authoritative suggestions. This skill enforces a strict evidence-based workflow that distinguishes between observations, approved requirements, and unknowns.
Use cases
- Onboarding agents to a complex codebase by providing a verified index of past decisions and architecture.
- Reconciling requirements when new stakeholder feedback conflicts with existing project documentation.
- Maintaining technical wikis by automatically linting for orphans and broken links after updates.
Known limitations
Does not perform mechanical enforcement of file permissions or external actions. Requires an initial schema and entrypoint to be established in the project area.
Requires an initial schema and entrypoint. Does not provide mechanical enforcement or permission for external actions. Knowledge remains a fallible view of evidence.
How to use
- 1
Connect your agent to the Agensi MCP
Copy the prompt and paste it into Claude, ChatGPT, Cursor or any agent that supports MCP. It sets up the connection for you. Manual setup
- 2
Sign in to Agensi
Your agent shows a sign-in link. Sign in and click Allow.
- 3
Ask for what you need
Your agent loads this skill and uses it on your task. It always asks before any payment.
Reviews
No reviews yet
Be one of the first to try it. Every listed skill goes through the trust checks below.
Security scanned
Checked by our 8-point scan before listing
Fresh listing
Recently published to Agensi
Money-back guarantee
Doesn't do what its page promises? You get your money back.
Trust & safety
Security scan
Security scan passed 14 days ago