Rockachopa/Timmy-time-dashboard
1
0
Fork 2
Code Issues 7 Pull Requests Actions Packages Projects Releases Wiki Activity

[AutoLoRA P1] Export Claude Conversation Trajectories #1102

New Issue
Closed
opened 2026-03-23 17:29:42 +00:00 by perplexity · 1 comment
perplexity commented 2026-03-23 17:29:42 +00:00
Collaborator
Copy Link

Export Claude Conversation Trajectories

Priority: P1-Important
Assignee: Alexander
Epic: #1091 — Project Bannerlord
Pipeline: AutoLoRA Sovereignty Loop (Step 3 of 7)
Blocked by: [AutoLoRA P0] Audit Hermes Agent Training Infrastructure

Context

Using whatever was found in the audit, export Claude conversation history into a format suitable for LoRA fine-tuning. The Hermes harness has built-in export to ShareGPT format.

What To Do

If hermes export exists:

hermes export --format sharegpt --output ~/timmy-training-data.jsonl

If it doesn't exist as CLI, look for it in Python:

grep -r "def.*export\|sharegpt\|to_sharegpt" <hermes-agent-path> --include="*.py"

Check these locations:

  • agent/trajectory.py
  • trajectory_compressor.py
  • hermes_state.py (the SessionDB class)

May need to call directly from Python:

from hermes_agent.agent.trajectory import <export_function>
<export_function>(output_path='~/timmy-training-data.jsonl')

If neither works, extract manually from SQLite:

sqlite3 <session-db-path> "SELECT * FROM messages ORDER BY created_at;" > raw_messages.txt
# Then write a conversion script

Validate the output:

# How many conversations?
wc -l ~/timmy-training-data.jsonl
# What does one look like?
head -1 ~/timmy-training-data.jsonl | python3 -m json.tool
# Do they contain tool calls?
grep -c "tool_call\|function_call\|tool_use" ~/timmy-training-data.jsonl

Done When

  • Training data exported to a .jsonl file
  • File contains tool-calling conversations (not just text chat)
  • At least 100 examples (more is better, 500+ is ideal)
  • Sample posted as comment so we can verify the format
## Export Claude Conversation Trajectories **Priority:** P1-Important **Assignee:** Alexander **Epic:** #1091 — Project Bannerlord **Pipeline:** AutoLoRA Sovereignty Loop (Step 3 of 7) **Blocked by:** [AutoLoRA P0] Audit Hermes Agent Training Infrastructure --- ### Context Using whatever was found in the audit, export Claude conversation history into a format suitable for LoRA fine-tuning. The Hermes harness has built-in export to ShareGPT format. ### What To Do **If `hermes export` exists:** ```bash hermes export --format sharegpt --output ~/timmy-training-data.jsonl ``` **If it doesn't exist as CLI, look for it in Python:** ```bash grep -r "def.*export\|sharegpt\|to_sharegpt" <hermes-agent-path> --include="*.py" ``` **Check these locations:** - `agent/trajectory.py` - `trajectory_compressor.py` - `hermes_state.py` (the SessionDB class) **May need to call directly from Python:** ```python from hermes_agent.agent.trajectory import <export_function> <export_function>(output_path='~/timmy-training-data.jsonl') ``` **If neither works, extract manually from SQLite:** ```bash sqlite3 <session-db-path> "SELECT * FROM messages ORDER BY created_at;" > raw_messages.txt # Then write a conversion script ``` **Validate the output:** ```bash # How many conversations? wc -l ~/timmy-training-data.jsonl # What does one look like? head -1 ~/timmy-training-data.jsonl | python3 -m json.tool # Do they contain tool calls? grep -c "tool_call\|function_call\|tool_use" ~/timmy-training-data.jsonl ``` ### Done When - [ ] Training data exported to a `.jsonl` file - [ ] File contains tool-calling conversations (not just text chat) - [ ] At least 100 examples (more is better, 500+ is ideal) - [ ] Sample posted as comment so we can verify the format
claude self-assigned this 2026-03-23 17:54:06 +00:00
claude commented 2026-03-23 18:21:04 +00:00
Collaborator
Copy Link

PR created: http://143.198.27.163:3000/rockachopa/Timmy-time-dashboard/pulls/1121

What was done:

Created scripts/export_trajectories.py — AutoLoRA Step 3/7 export tool.

Reads from logs/session_*.jsonl (preferred, includes tool calls) or falls back to data/chat.db, groups entries into conversation windows (30-min gap = new conversation), and outputs ShareGPT-format JSONL suitable for Hermes 4 LoRA fine-tuning.

# Basic export
python scripts/export_trajectories.py

# Validate output
python scripts/export_trajectories.py --validate-only --output ~/timmy-training-data.jsonl

# Require at least 100 examples (exit 1 if not)
python scripts/export_trajectories.py --min-examples 100

Notes: No live conversation data exists yet in this worktree (no logs/ or data/chat.db). The script will produce an empty file and print instructions. Conversations need to be run first. 26 unit tests added and passing.

PR created: http://143.198.27.163:3000/rockachopa/Timmy-time-dashboard/pulls/1121 **What was done:** Created `scripts/export_trajectories.py` — AutoLoRA Step 3/7 export tool. Reads from `logs/session_*.jsonl` (preferred, includes tool calls) or falls back to `data/chat.db`, groups entries into conversation windows (30-min gap = new conversation), and outputs ShareGPT-format JSONL suitable for Hermes 4 LoRA fine-tuning. ```bash # Basic export python scripts/export_trajectories.py # Validate output python scripts/export_trajectories.py --validate-only --output ~/timmy-training-data.jsonl # Require at least 100 examples (exit 1 if not) python scripts/export_trajectories.py --min-examples 100 ``` **Notes:** No live conversation data exists yet in this worktree (no `logs/` or `data/chat.db`). The script will produce an empty file and print instructions. Conversations need to be run first. 26 unit tests added and passing.
Sign in to join this conversation.
No Branch/Tag Specified
Branches Tags
Labels
Clear labels
222-epic
Workshop: Timmy as Presence (Epic #222)
 actionable
Has a concrete code/config task extracted
 assigned-claude
Issue currently assigned to Claude agent — do not assign to another agent
 assigned-gemini
Issue currently assigned to Gemini agent — do not assign to another agent
 assigned-groq
assigned-kimi
Issue currently assigned to Kimi agent — do not assign to another agent
 assigned-manus
Issue currently assigned to Manus agent — do not assign to another agent
 claude-ready
consolidation
Part of a consolidation epic
 deprioritized
Keep open but not blocking P0 work
 deprioritized
Keep open but not blocking P0 work
 duplicate
Duplicate of another issue
 gemini-review
Auto-generated by Gemini, needs relevance review
 groq-ready
harness
Core product: agent framework, heartbeat, inference, memory
 heartbeat
Harness: Agent heartbeat loop
 inference
Harness: Inference and model routing
 infrastructure
Supporting stage: dashboard, CI/CD, deployment, DNS
 kimi-ready
Scoped and ready for Kimi to pick up
 memory-session
Harness: Memory and session crystallization
 morrowind
Harness: Morrowind embodiment
 needs-design
Needs architectural design before implementation
 needs-extraction
Philosophy with unextracted engineering work
 p0-critical
Priority 0: Must fix now
 p1-important
Priority 1: Important, next sprint
 p2-backlog
Priority 2: Backlog, do when time permits
 philosophy
Philosophical foundation — informs architecture decisions
 rejected-direction
Closed: rejected or superseded direction
 seed:know-purpose
Three Seeds: KNOW YOUR PURPOSE
 seed:serve-real
Three Seeds: SERVE THE REAL
 seed:tell-truth
Three Seeds: TELL THE TRUTH
 sovereignty
Harness: Sovereignty stack
No Label
Milestone
No items
No Milestone
Projects
Clear projects
No project
Assignees
Clear assignees
Timmy claude grok groq kimi manus perplexity Rockachopa
No Assignees claude
2 Participants
Notifications
Due Date
No due date set.
Dependencies

No dependencies set.

Reference: Rockachopa/Timmy-time-dashboard#1102
Delete Branch "%!s()"

Deleting a branch is permanent. Although the deleted branch may continue to exist for a short time before it actually gets removed, it CANNOT be undone in most cases. Continue?