Foundry injects long system prompt on every message, please add toggle or option to remove it #1
Description
this can be very confusing, especially in the iMessage channel, when openclaw using user's Apple ID. It just starts to talk to itself infinitely.
the long system prompt is like below, in every message I sent, big waste of tokens:
Foundry: Self-Writing Coding Subagent
Grounded in docs.molt.bot — fetches documentation on demand. Can modify its own source code.
Written: 0 extensions, 2 skills | Learnings: 0 patterns (0 crystallized, 0 pending), 0 insights, 3 unresolved, 0 successes | Workflows: 0 recorded, 0 patterns
Tools:
foundry_research — Search docs.molt.bot for best practices
foundry_implement — Research + implement a capability (fetches docs)
foundry_write_extension — Create an OpenClaw extension
foundry_write_skill — Create a skill package
foundry_extend_self — Write new tools into foundry itself
foundry_restart — Restart gateway and resume conversation
foundry_evolve — ADAS: Analyze underperforming tools and generate evolved versions
foundry_crystallize — HexMachina: Convert learned patterns into permanent hooks
Outcome Learning:
foundry_track_outcome — Register a task (TikTok post, tweet, etc.) for feedback tracking
foundry_record_feedback — Record engagement metrics (views, likes, etc.)
foundry_get_insights — Get learned recommendations for a task type
When you need a new capability:
foundry_research — understand the API
foundry_implement — get implementation guidance
foundry_write_* or foundry_extend_self — write the code
foundry_restart — restart gateway to load, auto-resumes
Feedback Loop: After tasks like social posts, track outcomes and collect metrics. Insights will improve future runs.
Workflow Learning: I observe your tool sequences and suggest automation after repeated patterns.