docs: add persistent memory guide#641
Open
hieptl wants to merge 3 commits into
Open
Conversation
5 tasks
|
Preview deployment for your docs. Learn more about Mintlify Previews.
💡 Tip: Enable Workflows to automatically generate PRs for you. |
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Summary of changes
Companion docs for OpenHands/software-agent-sdk#4178 (SDK issue #2037), which adds opt-in persistent memory to the SDK. Review on that PR asked for an example and docs: the example ships there (
examples/01_standalone_sdk/55_persistent_memory.py); this PR adds the documentation.sdk/guides/persistent-memory.mdx— covers enabling the feature (AgentContext(load_memory=True), opt-in and off by default); the two memory tiers (user~/.openhands/memory/, project<workspace>/.openhands/memory/); the file model (MEMORY.mdcurated index injected at session start vs. daily logs read on demand); what gets injected (the<MEMORY_CONTEXT>block,<UNTRUSTED_CONTENT>wrapping, ~6,000-character truncate-from-top budget, user tier before project tier); operational properties (re-resolved each session, excluded frombase_state.jsonand API payloads, best-effort loading); and the maintenance habits the system prompt teaches the agent. Ends with the standard Ready-to-run Example section (<RunExampleCode>+ GitHub link) embeddingexamples/01_standalone_sdk/55_persistent_memory.py.sdk/guides/persistent-memoryindocs.jsonunder SDK → Guides, right after Context Condenser.Verification
.github/scripts/sync_code_blocks.pyextract_code_blocks+normalize_contentcomparison).docs.jsonre-parses as valid JSON after the nav edit./sdk/guides/skill,/sdk/guides/context-condenser,/sdk/guides/convo-persistence, shared snippet/sdk/shared-snippets/how-to-run-example.mdx), and the#ready-to-run-exampleanchor matches its heading.Notes
main, and thesync-docs-code-blocksworkflow reads example content from the SDK repo'smain. Until then the sync workflow logs a missing-source warning for this block and skips it (not a CI failure), and the PR-levelcheck-broken-linksjob only checks internal links, so CI on this PR is unaffected.llms.txt/llms-full.txtare intentionally untouched: regenerating now bundles unrelated drift from previously merged pages (e.g. the Fork-a-Conversation guide), and the scheduledcheck-llms-files.ymlworkflow owns that sync.