Memory

a.Externalizing state

Externalize state beyond the context window

Nothing lives here yet.

b.Resuming

Recover and resume rather than restart

1. i.

   Filesystem-based agent state and checkpoints

   Save the agent's state at each step boundary (the conversation, the workspace files, process state when it matters, and a ledger of external side effects) so a crashed or stuck run resumes or rolls back instead of starting over.

   Better qualityLower costMixed performance

   Tier 2 · reliability · changes state · any agent · 6 sources, 4 measured

2. ii.

   Initializer-maintainer dual agent

   Give the first session of a multi-session build a setup-only prompt (expand the spec into a JSON list of end-to-end tests, all failing, and write an init script, a progress file and a first commit) and every later session a maintainer prompt that reads that state, re-tests what passed, builds one failing item, verifies it end to end and commits.

   Better qualityBetter governanceMixed costSlower

   Tier 4 · restructuring · changes state · planning, code generation agents · 6 sources, 1 measured

3. iii.

   Keep plan, progress and results in files when work outlives one context window

   When a task spans context resets (compaction, a truncation limit, a new session), have the agent keep what the next window needs in files and read them back at the start of each window: the plan and a progress list, decisions and open items, intermediate results too large to hold in context, and for coding a structured test list and an init script.

   Better qualityBetter governanceMixed costMixed performance

   Tier 4 · restructuring · changes state · planning agents · 15 sources, 9 measured

c.Reusing work

Reuse prior work across runs

1. i.

   Episodic memory retrieval

   Log every agent session as a raw trace, distill a few verified lessons from them (user corrections, definitions, workflows that worked, mistakes) into short memory files, show the agent an index of those files at the start of each task, and let it search both.

   Better for brandMixed qualityMixed costMixed performance

   Tier 4 · restructuring · changes state · memory agents · 9 sources, 7 measured

2. ii.

   Skill library evolution

   Have the agent turn runs that passed a verifier into reusable skills (a code function or a SKILL.md procedure), store them in a library, and on later tasks show it only the few a separate selection step picks, reviewing what goes in and pruning what goes unused.

   Mixed qualityMixed costMixed performance

   Tier 4 · restructuring · changes state · memory agents · 6 sources, 4 measured