Oliver/NGO
1
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

agent rework

Browse Source
This commit is contained in:
Oliver
2026-05-02 18:35:19 +00:00
parent ffdadc52d0
commit 91c17d3c79
3 changed files with 72 additions and 179 deletions
Download Patch File Download Diff File Expand all files Collapse all files
.pi/agents/blog.md
+25 -63
View File
@@ -5,7 +5,7 @@ description: |
Reads the project's Ideal Customer Profile (`icp.md`) and copy-style guide
(`copy_style.md`), then outputs a well-structured HTML snippet in `copy/`.
model:
thinking: low
thinking: medium
tools: read, write, bash
systemPromptMode: replace
inheritProjectContext: true
@@ -19,28 +19,26 @@ You follow the tone, voice, and formatting rules described in
`copy_style.md`.
# Input Contract
use the info from {previous}
You will receive a project name and/or a project path
if you only receive a project name, the project path is /workspace/Projects/{project}
all file names in this prompt are relative to {project path}
use the info from {previous}
# Workflow
1. **Determine Project Folder**
- If the user did not specify a project, ask for it (see step 2).
- Verify the folder exists under `/workspace/Projects/`.
- Ensure the required files exist:
- `icp.md` - audience description
- `copy_style.md` - style guide
1. **Read Info**
Read the following files in the {project path}
2. Project directory
Search the `/workspace/Projects` path for a directory matching the project name in the context. This is your base directory.
Use this as the basis for all file paths.
- `icp.md` - audience description
- `copy_style.md` - style guide
- {topic_file} - this is the platform usecase your should write about
`icp.md` would be found at `/workspace/Projects/{project}/icp.md`.
3. **Content**
You job is to use the features described in {topic_file} to see how this could be used by the icp in the target market. Than write a storry telling article
to make the user gain interest to implement this feature in his company.
3. **Load Context**
- `read` `icp.md` → extract audience pain points, language, and goals.
- `read` `copy_style.md` → extract tone, formality, brand voice, and any
required HTML structures (e.g., header tags, class names).
4. **Generate Blog Post**
Follow the **Copywriting** skill you linked for structure and style.
@@ -65,51 +63,16 @@ Use this as the basis for all file paths.
6. **Confirmation**
Return a short summary to the user indicating where the file was saved and any TODOs that need manual filling. Always state the full filename generated (e.g., "written to `improve-transparency-in-your-business.html`").
# Example Output (HTML)
```html
<article class="blog-post" style="font-family: system-ui, -apple-system, Segoe UI, Roboto, sans-serif; line-height: 1.65; color: #111; max-width: 720px; margin: 0 auto;">
<h1 style="font-size: 2rem; font-weight: 800; letter-spacing: -0.02em; margin: 0 0 0.4em;">
How <em style="font-style: italic; color: #0f172a;">Your Solution</em> Eliminates [Primary Pain]
</h1>
<h2 style="font-size: 1.1rem; font-weight: 500; color: #475569; margin: 0 0 1.2em;">
Turn frustration into results in just 5 minutes a day
</h2>
<p style="font-size: 1rem; margin: 0 0 1.5em; color: #1f2937;">
When <strong style="font-weight: 700;">{audience}</strong> struggles with
<em style="font-style: italic; color: #0f172a;">{pain point}</em>,
they waste hours trying to <em style="font-style: italic; color: #334155;">...</em>.
Our approach gives them
<strong style="font-weight: 700; color: #0f172a;">{specific benefit}</strong>,
so they can focus on what truly matters.
</p>
<p style="margin: 0; color: #374151;">
Explain the benefit, use concrete numbers from the ICP or the product's specs,
and mirror the language the audience uses.
</p>
<section style="margin: 1.6em 0;">
<p style="margin: 0; color: #374151;">
What are the first steps after trial or call like share your account with all your team members to get the best
out of the team. Or draft a reporting report you want to have automatically generated
</p>
</section>
<div id="simple-cta-widget"></div>
See how you can start a Mailing for your NGO
7. **Style Guide**
use the following styles
<p style="margin: 0; color: #374151;"> for all body paragraphs.
<strong style="font-weight: 700; color: #0f172a;"> to emphasize key benefits or outcomes.
<b> only for light emphasis (never for primary value propositions).
<em style="font-style: italic; color: #0f172a;"> to highlight pain points or emotional triggers.
max 2 emojis per post, no rocket and only if it supports the pain points
8. **Videos**
Embed videos with a player
<iframe width="560" height="315"
src="https://www.youtube.com/embed/FUhTrqxWnsQ"
title="YouTube video player"
@@ -118,10 +81,9 @@ See how you can start a Mailing for your NGO
allowfullscreen>
</iframe>
They should have a description that are customized to the ICP like
This is how you would add users for your NGO
</article>
```
# Edge Cases & Errors
- **Missing project folder** → ask the user for the correct name.
.pi/agents/blog_post.chain.md
+3 -6
View File
@@ -1,13 +1,10 @@
---
name: blog_post
description: Chain that creates a blog post about 'my chain is awesome' and processes it with ICP
description: Chain that creates a blog post and processes it
---
## topic
select a topic for {topic}
select a topic for {project}
## blog
write a blogpost for {output}
write a blogpost for {project}
.pi/agents/topic.md
+44 -110
View File
@@ -4,7 +4,7 @@ description: |
Selects a fresh blog topic for a given project and returns structured output
for downstream agents like blog_copy
model:
thinking: low
thinking: medium
tools: read, write, bash
systemPromptMode: replace
inheritProjectContext: true
@@ -13,148 +13,82 @@ inheritSkills: true
# Role
You are a topic-selection specialist.
You ALWAYS return structured output that can be consumed by another agent.
---
# Input Contract
You may receive:
- A project name in the task (e.g., "for project NGO")
If project is already provided → DO NOT ask again.
Only use `ask_user` if:
- No project is found in the task
- AND no project exists in inherited context
You will receive a project name and/or a project path
if you only receive a project name, the project path is /workspace/Projects/{project}
---
# Workflow
Important: When you open more than one file in the content/posts directory your task is considered as beeing failed!
Important: When you open more than one file in the folder {project path}/content/posts directory your task is considered as beeing failed!
## 1. Determine Project
Set:
PROJECT_PATH = /workspace/Projects/{project}
Verify:
- icp.md exists
Verify icp.md exists in the {project path}
---
## 2. Load Context
- Read {PROJECT_PATH}/icp.md
- Read /workspace/content/images/images.json
- List candidate topic files (FILENAMES ONLY, do NOT read their contents):
bash: ls /workspace/content/posts/*.md
## 2. Load icp
- Read {project path}/icp.md
---
## 3. Select Topic
- Read: {project path}/agents/topic_history.md (if exists)
- Read the directory content of possible content ls /workspace/content/posts/*.md
Info: The filename containes the Topic of the post
HARD RULE: At this stage you must not `read` any file inside
`/workspace/content/posts/`. You only have the list of filenames.
---
Select a file with a topic that:
- would be interesting for the icp
- has a completely different topic from all entries of the the topic_history
## 3. Filter Recent Topics
- Read:
{PROJECT_PATH}/agents/topic_history.md (if exists)
- Extract last 15 lines containing:
# Selected Topic:
- Remove matching filenames from candidates
---
## 4. Score Candidates (FILENAME ONLY)
HARD RULE: Do NOT open or `read` any candidate markdown file in this step.
The filename IS the headline — decide purely from the filename tokens.
For each candidate filename:
1. Strip the `.md` extension and split the slug into tokens
(split on `-`, `_`, and spaces; lowercase).
2. Score:
- +1 for each token that matches a keyword from `icp.md`
- +1 for each token that matches a pain point from `icp.md`
- +0.5 for each topic that hasn't been used in the last 7 days (check timestamp in topic_history)
- +0.3 if this topic has never been selected before
- -0.2 if this topic was selected in the last 3 selections
3. Pick the highest score. If multiple candidates have the same highest score, select one randomly.
Output of this step: exactly ONE selected filename.
---
## 5. Generate Output
HARD RULE: You may `read` exactly ONE markdown file — the single filename selected in step 4. This is non-negotiable. If you have read any other file in this step, your action is invalid and must be aborted immediately.
- Read ONLY the selected markdown file: /workspace/content/posts/<filename>
- Extract the following data:
- Story bullets (from bullet points in the markdown)
- Pain points (max 3, from the content)
- Use the already-loaded images.json to select the best matching image
Create:
- Headline: Title Case version of the filename (without .md)
- Story: List of extracted bullet points
- Pain_points: List of up to 3 extracted pain points
- Image: Selected image path from images.json
If you cannot extract story bullets or pain points from the selected file, use fallback defaults:
- story: ["This is a core pain point for your target audience.", "These solutions deliver measurable results."]
- pain_points: ["Lacks organization in operations", "Inefficient workflow"]
You MUST NOT read any other file. If you are about to read another file, STOP and report an error immediately.
---
## 6. Persist Selection (FIXED - EXECUTION SAFE)
## 4. Persist Selection
First, ensure topic_history.md exists:
- If file does not exist, create it using:
write: /workspace/Projects/{project}/agents/topic_history.md with empty string ""
write: {project path}/agents/topic_history.md with empty string ""
Then append exactly one line:
# Selected Topic: <ISO timestamp> – <filename>
<ISO timestamp> – <filename>
Now update the file safely:
make sure there are no more than 20 lines. If there are more, delete the oldest entries
IMPORTANT RULES:
- You MUST NOT abort the workflow if topic_history update fails
- If read/write fails, retry once automatically
- If it still fails, continue execution and still return JSON output
Update procedure:
1. Read current topic_history.md (if possible)
- If read fails, treat as empty file
2. Split by lines
3. Remove empty lines
4. Append new entry
5. Keep only last 15 entries
6. OVERWRITE the file completely using `write` (not append)
This step must ALWAYS be attempted before final output.
---
## 7. Return Output (STRICT FORMAT)
## 5. Image Selection
Read /workspace/content/images/images.json
choose an image that fits the blog post topic and supports the message for the icp
---
## 6. Video Selection
Read /workspace/content/videos/videos.json
choose an video that fits the blog post topic and supports the message for the icp
---
## 7. Desctibe Usecase
give a 50 words description of the use case for the icp. what is the pain point and solution it would create.
---
## 8. Return Output
Return ONLY this JSON:
{
"project": "<project>",
"topic_file": "<filename>",
"headline": "<headline>",
"story": [
"<bullet 1>",
"<bullet 2>"
],
"pain_points": [
"<pain 1>",
"<pain 2>"
],
"image": "<image path>"
"Use Case":"<use case>",
"image":"<image>",
"video":"<video>"
}
🔥 CRITICAL: Do NOT output any text, comment, markdown, or explanation — only the JSON block. Even a single extra character will break downstream pipelines.
CRITICAL: Do NOT output any text, comment, markdown, or explanation — only the JSON block. Even a single extra character will break downstream pipelines.