System-Level Proxy
Intercepts all Claude traffic at the OS level — no manual env vars needed.
Layer 1 — Deterministic Rules Always On
Fast regex + entropy detection. Runs on every request, zero latency.
Redaction Replacement
Replace detected sensitive text with your own label, such as [PRIVATE]. Leave blank to use [REDACTED_EMAIL], [REDACTED_SECRET], and other typed defaults. Applies to deterministic rules.
Action Mode
Layer 2 — LLM Reviewer Optional
Local open-source model via Ollama catches context-aware leaks that regex misses.
Audit Log
Prompt Recordings
Connectors
Configure credentials once — all linked tools will use them automatically.
Tools
Select which tools to expose in your MCP client.
Clients
Choose your AI client. We'll write the MCP config automatically.
Agent Memory
Persistent memory so AI agents can hand off work across sessions.
Checkpoints
Strategy
Free-form markdown. Agents read this to understand your project goals and constraints.
Stored Facts
Session Log
Troubleshoot
Common issues and fixes.
401 Bad Credentials
The GitHub API rejected your token — expired, wrong scopes, or pasted incorrectly.
Fine-grained PATs hit unexpected 401s for org repos. A Classic PAT with repo scope is most reliable.
- Go to github.com/settings/tokens/new
- Note:
agentic-store-mcp - Scopes: check repo
- Click Generate token and copy immediately
Go to → GitHub → paste token → Save → Test Connection.
Secrets configured in webapp not seen by MCP server
Webapp and MCP server run in separate containers — both must mount the same volume.
docker compose up -d
Go to and save your tokens.
Your client config must include the -v flag. See the tab for the snippet.
Tool returns "PyGithub not installed" or import error
Dependencies aren't installed yet.
uv sync
Server disconnected — "No module named 'mcp'"
MCP server exited immediately. Virtual environment is missing or stale.
cd /path/to/agentic-store-mcp uv sync
deactivate uv sync
Restart Claude Desktop / Cursor so the MCP server relaunches with the updated environment.
Relay
Jira for AI Agents. Plan work, coordinate agents, and track delivery.