11---
22name : scholar
33description : Self-Evolving Research Agent — scans web for relevant tech, evaluates findings, proposes improvements. Uses Perplexity Sonar API via MCP.
4- argument-hint : [scan|eval|apply|full|topic:"query"]
4+ argument-hint : [scan|eval|apply|full|report| topic:"query"]
55allowed-tools : Bash(gh *), Bash(cat *), Bash(grep *), Bash(find *), Bash(python3 *), Bash(echo *), Bash(date *), Bash(wc *), Bash(git *), Bash(test *), Bash(ls *), Read, Edit, Write, mcp__perplexity__perplexity_search, mcp__perplexity__perplexity_ask, mcp__perplexity__perplexity_research, mcp__perplexity__perplexity_reason
66---
77
@@ -19,6 +19,7 @@ Parse $ARGUMENTS to determine mode:
1919- ` eval ` — Run EVAL phase on last scan results
2020- ` apply ` — Run APPLY phase (create issues / enrich MU)
2121- ` full ` — Run all 3 phases sequentially (default if no args)
22+ - ` report ` — Show last scan results without running new scan (for bridge/Perplexity)
2223- ` topic:"<query>" ` — Deep research on a specific topic
2324- ` errors ` — Scan for solutions to current broken specs/compilation errors
2425- ` zig ` — Scan for Zig 0.15 updates and best practices
@@ -256,19 +257,31 @@ Render a report after each run:
256257 ✨ Scholar says: "{contextual insight about findings}"
257258```
258259
260+ ## Mode: report
261+
262+ If $ARGUMENTS is ` report ` , do NOT run any scan. Instead:
263+
264+ 1 . Read the latest scan file from ` .trinity/scholar/ ` (most recent ` scan_*.json ` )
265+ 2 . Render the Output Format report above using cached data
266+ 3 . If no scan file exists, output: "No scan data. Run: /scholar scan"
267+
268+ This mode is optimized for bridge-agent / Perplexity queries — fast, no API calls.
269+
259270## Cron Integration
260271
261272Scholar can be triggered remotely via bridge-agent:
262273```
263274claude:Run /scholar full
264275claude:Run /scholar errors
276+ claude:Run /scholar report
265277claude:Run /scholar topic:"ternary neural network quantization"
266278```
267279
268- For automated 24h cycle, add to bridge cron:
269- - 08:00 — ` claude:Run /scholar zig ` (morning: language updates)
270- - 20:00 — ` claude:Run /scholar errors ` (evening: fix broken specs)
271- - 02:00 — ` claude:Run /scholar full ` (night: deep research)
280+ ### Automated 24h cycle (via bridge-agent cron):
281+ - ** 06:00 UTC** — ` claude:Run /scholar full ` (morning: full scan + eval + apply)
282+ - ** 18:00 UTC** — ` claude:Run /scholar errors ` (evening: fix broken specs)
283+
284+ The bridge-agent checks UTC hour and auto-submits scholar jobs.
272285
273286## Translation Table (EN → RU)
274287
0 commit comments