@dreadedzombie/pi-init 1.0.0 → 1.2.0

This diff represents the content of publicly available package versions that have been released to one of the supported registries. The information contained in this diff is provided for informational purposes only and reflects changes between package versions as they appear in their respective public registries.
package/README.md CHANGED
@@ -5,10 +5,14 @@ Adds `/init` to [Pi](https://pi.dev/) — generates or updates an AGENTS.md in y
5
5
  ## Install
6
6
 
7
7
  ```bash
8
+ # via npm (recommended)
9
+ pi install npm:@dreadedzombie/pi-init
10
+
11
+ # via GitHub
8
12
  pi install https://github.com/joenilan/pi-init
9
13
  ```
10
14
 
11
- Pi clones the repo and loads the extension automatically on next start.
15
+ Pi loads the extension automatically on next start.
12
16
 
13
17
  ## Usage
14
18
 
package/package.json CHANGED
@@ -1,6 +1,6 @@
1
1
  {
2
2
  "name": "@dreadedzombie/pi-init",
3
- "version": "1.0.0",
3
+ "version": "1.2.0",
4
4
  "description": "Generates a typed AGENTS.md for your project — /init, /init research, /init debug, /init code",
5
5
  "license": "MIT",
6
6
  "author": "DreadedZombie",
package/src/index.ts CHANGED
@@ -60,12 +60,23 @@ const TEMPLATE_RESEARCH = `# Research Agent
60
60
 
61
61
  **Started:** <!-- Pi: insert today's date (YYYY-MM-DD) -->
62
62
 
63
+ ## FIRST ACTION — Before anything else
64
+ Create the \`research/\` directory and write \`research/plan.md\` listing the topics you will investigate.
65
+ Do this before any searching or reading. This is not optional.
66
+
67
+ ## Output Rule — Files Before Chat
68
+ **Never summarize findings in chat.** Every finding goes into a file.
69
+ - Create \`research/<topic>.md\` as you complete each topic — not at the end
70
+ - Update the Research Findings section in this file after writing each doc
71
+ - Chat response comes last, only to say what files were written and what to do next
72
+ - **If you are about to type a finding into chat — stop. Write it to a file instead.**
73
+ - The goal is a handoff document set for the next agent, not a conversation
74
+
63
75
  ## Search Protocol
64
76
  - Run at least **4 distinct queries** approaching the topic from different angles before drawing conclusions
65
77
  - Read the **full content** of at least 6 sources — not just summaries or snippets
66
78
  - For JS-heavy or dynamically rendered pages where \`fetch\` returns sparse HTML, use **Playwright** to render and extract
67
79
  - Cross-reference findings across sources before responding
68
- - Do **not** respond until you have read multiple sources and built a complete picture
69
80
 
70
81
  ## Source Quality Standards
71
82
  - Prefer primary sources: official documentation, research papers, original announcements
@@ -78,9 +89,18 @@ const TEMPLATE_RESEARCH = `# Research Agent
78
89
  2. **Dive** — read the full content of the 5-6 most relevant sources
79
90
  3. **Cross-reference** — identify what sources agree and disagree on
80
91
  4. **Synthesize** — build a coherent picture with citations
81
- 5. **Save** — write findings to \`research/<topic>.md\`, one file per topic or question
82
- 6. **Update** — add each saved file to the Research Findings section below with a one-line summary
83
- 7. **Gaps** explicitly note what you could not find or verify
92
+ 5. **Write** — save to \`research/<topic>.md\` with full detail, citations, code snippets, and gaps
93
+ 6. **Update AGENTS.md** — add the file to Research Findings below, one line per file
94
+ 7. Repeat steps 1-6 for each sub-topic before moving on
95
+
96
+ ## research/<topic>.md format
97
+ Each file should contain:
98
+ - **What was found** — detailed findings with citations (URLs)
99
+ - **Key facts** — bullet list of the most important discoveries
100
+ - **Conflicts** — where sources disagreed and which is more credible
101
+ - **Code / config** — any relevant snippets, commands, or configuration found
102
+ - **Gaps** — what could not be found or verified
103
+ - **Next steps** — what a follow-up agent should investigate or try
84
104
 
85
105
  ## Tools
86
106
  - \`searxng\` — start here for all searches (private, no tracking)
@@ -91,9 +111,9 @@ const TEMPLATE_RESEARCH = `# Research Agent
91
111
  - \`pi-docparser\` — for PDFs, papers, Word docs, spreadsheets
92
112
 
93
113
  ## Research Findings
94
- <!-- Pi: as you complete each topic, save findings to research/<topic>.md and add a line here:
114
+ <!-- Pi: after writing each research/<topic>.md, add a line here immediately:
95
115
  - [Topic title](research/filename.md) — one-line summary of what was found
96
- Update this section incrementally — do not wait until the end. -->
116
+ Do not wait until the end. Update this section after every file written. -->
97
117
  `;
98
118
 
99
119
  const TEMPLATE_DEBUG = `# Debug Agent