Palinode Session Memory

This skill keeps your AI agent's memory fresh across coding sessions using Palinode MCP tools.

On Session Start

Search for prior context before beginning work:

palinode_search(query="[current project or task description]", limit=5)

Review results and reference relevant decisions or blockers from previous sessions.

During Work — Save Milestones

After each major milestone, save the outcome:

palinode_save(
  content="[what was accomplished and why]",
  type="Decision",          # or "Insight" for reusable lessons
  project="[project-slug]"
)

When to save:

• Tests pass after a significant change
• Feature is complete and working
• Architectural or design decision made (include rationale)
• Bug fixed that took >15 minutes (save the root cause)
• Something surprising discovered (save as Insight)

When NOT to save:

• Routine file edits, typo fixes
• Intermediate debug steps (save the resolution only)
• Things git already tracks (code changes, file history)

Every ~30 Minutes

If actively working and 30+ minutes since last palinode_save, save a brief progress note:

palinode_save(
  content="Progress: [what's been done so far, what's next]",
  type="ProjectSnapshot"
)

On Session End

Before the user exits, capture the session:

palinode_session_end(
  summary="[1-2 sentence summary of accomplishments]",
  decisions=["decision 1 with rationale", "decision 2"],
  blockers=["open question or next step"],
  project="[project-slug]"
)

Tool Reference

Tool	When
palinode_search	Start of session, or "what do we know about X"
palinode_save	Milestones, decisions, insights, progress
palinode_session_end	End of session — structured summary
palinode_diff	"What changed recently?"
palinode_blame	"When was this decided?"
palinode_trigger	Register auto-recall for recurring topics