"Health check for prism-mcp. Reports dashboard reachability, MCP server connection, knowledge base population, and version."
Install
npx skillscat add jthorare/agentic-workflow/prismstatus Install via the SkillsCat registry.
Prism Status — Verify Mind Palace Health
Quick health check for the prism-mcp Mind Palace.
Before anything else: read $HOME/.agentic-workflow/toolkit/skills/_preamble.md and follow it (skill index, provider capability map, bootstrap check, session context). Run its Session Close section when this skill finishes.
Preamble exemption: If the preamble's prism-mcp calls failed, continue anyway — diagnosing that failure is this skill's purpose; Session Close ledger/handoff are exempt for this skill.
Overview
Probes the prism-mcp dashboard HTTP endpoint, the MCP server connection, and the per-repo knowledge base. Returns a single status report; writes no file. Useful at session start (matches the prism-context.sh SessionStart hook) and on demand.
Inputs
--port <port>(optional) — overridePRISM_DASHBOARD_PORT. Default port comes from the env var, or7180if unset.
Steps
Assign PORT explicitly at the top of every bash block (shell state does not persist between blocks): --port arg if provided, else $PRISM_DASHBOARD_PORT, else 7180.
Dashboard reachability:
PORT="${PRISM_DASHBOARD_PORT:-7180}" # replace with the --port value if one was given curl -fsS --max-time 3 -o /dev/null "http://localhost:$PORT/"Capture exit code and body. Report ✓ on 0, ✗ on non-zero.
MCP tool surface:
Callmcp: prism-mcp/session_health_check. Report ✓ on success, ✗ + error on failure.Knowledge base population for current repo:
Resolve the repo slug via the shared helper:SHARED_DIR="$HOME/.agentic-workflow/toolkit/skills/_shared" source "$SHARED_DIR/repo-slug.sh" echo "repo-slug: $REPO_SLUG"Call
mcp: prism-mcp/session_load_contextwithproject=$REPO_SLUGandlevel="minimal". Report ✓ "KB has N entries" or ✗ "KB empty".Version — structured sources, in order of preference:
- The
versionfield of thesession_health_checkresponse from check 2, if present. - The dashboard
/healthJSON body:PORT="${PRISM_DASHBOARD_PORT:-7180}" # replace with the --port value if one was given curl -fsS --max-time 3 "http://localhost:$PORT/health" | jq -r '.version // "unknown"' - Fallback: the host CLI's MCP registration for
prism-mcp, taking the first strict semver match ([0-9]+.[0-9]+.[0-9]+) —claude mcp get prism-mcp(Claude Code),codex mcp get prism-mcp(Codex), or theprism-mcpentry in~/.cursor/mcp.json(Cursor):{ command -v claude >/dev/null && claude mcp get prism-mcp 2>/dev/null; \ command -v codex >/dev/null && codex mcp get prism-mcp 2>/dev/null; \ jq -c '.mcpServers["prism-mcp"] // empty' "$HOME/.cursor/mcp.json" 2>/dev/null; } \ | grep -oE '[0-9]+\.[0-9]+\.[0-9]+' | head -1
Report the version (e.g., "5.1.0"); "unknown" if all three fail.
- The
Summarize with a 4-line block — one line per check (dashboard, MCP surface, KB, version) — each prefixed
✓or✗. If any check failed, add a fix suggestion line at the bottom (re-run setup.sh, check env, etc.).
Outputs
stdout only. No file written.
Next steps
/officeHours— if KB is empty, start populating it with a brainstorm/bootstrap— if you've never run it on this repo, do so to seed planning docs