# Changelog [English](CHANGELOG.md) | [한국어](CHANGELOG.ko.md) All notable changes to `pi-memnest` are recorded here. ## [0.6.0] - 2026-11-14 ### Security - Task cards are written through the core's `POST /task-card` when the core reserves the card collections (it reports this in `/health`). Saving, recording user messages, finishing and reopening work as before. With an older core the extension keeps writing through `/add` and `/update`. - Card titles are cut to 120 characters on one line with `<` and `>` replaced, the card list says that titles are names and not instructions, and card text put into the prompt cannot close the reminder wrapper. - Vault tools send the core's API token. Without `MEMNEST_TOKEN`, the extension reads `~/.memnest/api-token`, or `MEMNEST_DATA_DIR/api-token` when that variable is set, and sends it only when `MEMNEST_URL` is a loopback address. The path is never taken from the server's reply. With `MEMNEST_TOKEN` set for a core that has none, the vault falls back to the file token. When the token cannot be read and the core asks for it, the tool returns an error that says so. - A bearer token is never sent over plain HTTP to an address that is not loopback; the request is refused before it leaves. ### Added - `/memnest-recent` lists the 20 newest saved memories (not transcript turns) and shows one in full. `/memnest-review` goes through saved rules and preferences, longest unconfirmed first, and asks for each one: Keep (records a review date), Delete (to the trash, restorable for 30 days; only on that choice), Skip, or Stop, then prints how many were kept, deleted and skipped. Both take `project:` or `project:all`, print text without a UI, and need a core with `POST /recent` and `POST /reviewed`; an older core gets a one-line "too old" message. - `/memnest-cards` and `/memnest-search`, run from the pi terminal UI without the model. `/memnest-cards` lists active and finished task cards (newest first, last save in local time, `[open]` for the cards this session records into), shows a card with the user's own words, and opens, closes or marks done the chosen card (done asks first). `/memnest-search [project:|project:all] ` runs the same search as `memory_search`, shows up to 10 results with project, date, type, score and excerpt plus the scope hint, and shows a chosen result in full; project names complete after `project:`. Without a UI they print text; with the core down or too old they say so in one line. - `memory_search` shows the core's scope hint. When a search of the current workspace missed clearly stronger matches in other projects, the result ends with one line that names those projects and says to search again with `project=""` or `project="all"`. Needs a core that sends `scope_hint`; with an older core nothing changes. - Saving a new task card passes on the core's list of rules and preferences already stored for that work, so they can be copied into the card. Needs a core that reserves the card collections. - `memory_remember` shows the core's list of existing memories that may cover the same subject (`same_subject` and `same_subject_note` in the reply), and `memory_update` takes `supersedes=` so the new memory can replace an old one without being saved again. Needs a core that sends the list; with an older core nothing changes. ### Changed - Automatic recall (off by default) asks the core's `POST /context`, which returns only saved memories on the prompt's topic, instead of taking `/search` results above a score. With the real extension against a test core holding 12 saved memories: three vague prompts and one unrelated prompt got no card (the score threshold put one memory into each of the three vague ones), and a Korean, a mixed Korean-English and an English question each got exactly their memory (the score threshold added an unrelated one to two of them). `MEMNEST_AUTOCONTEXT_MIN_SCORE` is used only with a core whose `/context` does not filter by topic, where recall works as before through `/search`. `MEMNEST_AUTOCONTEXT_N` is now the most memories asked for (the core returns at most 20). The other settings and limits are unchanged, and no setting was removed. - The viewer the `/memnest-...` commands show long text in names its scroll keys in the title: the arrow keys and Ctrl+PageUp/Ctrl+PageDown. In pi's fullscreen terminal mode plain PageUp and PageDown scroll the transcript, not this view. ## [0.5.0] - 2026-11-08 ### Added - Task cards, on by default. They add the `task_card` tool to save, list, open, and finish handoff cards for work that spans sessions. Active cards live in a shared `task-cards` collection and finished ones in `task-cards-done`. Saving the same title supersedes the previous card. The extension appends every user message to the open card verbatim, outside the model's summary. Up to three opened cards are pinned into the system prompt on every run with a fixed user-policy-first line; the pinned text is re-rendered only when first pinned and after a compaction, so it does not break the prompt cache between prompts. A new session lists up to 20 active titles ranked by its first request; it says when to save a new card, including when other cards exist. A user message is written into the open card when the run settles, so a card opened by mistake or closed because the user moved on does not keep it. The first prompt after a compaction asks the model to save what changed; with no card open, it reminds once per session to save one if the work continues later. The first `memory_remember` of a session with no card open gets one line asking for a card when the fact belongs to work that spans sessions. A card open but not saved for 8 user messages, counted across a resume, gets one reminder to close it if the conversation moved on, or save it. `/memnest` lists the active cards with their last save in local time and the cards this session records into. Measurement and limits are in `docs/task-cards.md`. ### Compatibility and limits - The default surface grows to the five memory tools plus `task_card`, and one prompt hook. It lists cards on a session's first prompt and pins an opened card. Autocontext stays opt-in. - Cards are found through hybrid search on a marker word, at most 50 per query; a topic search reaches cards beyond the newest 50. Session-start cost grows with active cards (about 0.43 s at 1,020 active, 3 ms at 20 active plus 1,000 finished), and lists with 50 or more active cards ask the model to offer cleanup. ## [0.4.0] - 2026-10-07 ### Added - `memory_remember(code_files=[...])` explicitly captures bounded file fingerprints through the matching core CLI, without storing file contents. - `memory_get(check_code=true)` explicitly compares a saved baseline in the current workspace, including remote-store baselines through a network-free local compare command. - `memory_search(evidence_only=true)` forwards optional local-model source selection and preserves its unverified assertion. - Optional `MEMNEST_ALLOWED_MODEL_PROVIDERS` blocks new memory/credential reads and automatic recall for unapproved or unknown model contexts before network access. ### Compatibility and limits - Five memory tools remain the default surface; automatic recall and code checks remain opt-in. - Provider allowlisting is a trusted-client routing safeguard, not endpoint attestation, retroactive history erasure, or a complete DLP boundary. - New source/evidence options require core 0.4.0. Existing basic operations remain compatible. ## [0.3.1] - 2026-09-28 ### Added - Show the core's optional `following_id` in search results with a warning that it is a later assistant capture, not a verified reply. Read that source with the existing `memory_get` tool. - Keep compatibility with cores that omit the hint; existing result excerpts and search limits are unchanged. This release also includes the 0.3.0 changes below, which were not previously published to npm. ## [0.3.0] - 2026-09-15 ### Changed - Autocontext is off by default. Only an explicit `balanced` or `aggressive` mode enables it; unrecognized modes fail closed. The five memory tools and conversation capture remain available independently. - `/memnest` reports the pi Autocontext state. - Opt-in memory cards include ID, project, and creation time for source verification. Missing IDs/projects and missing or nonnumeric/nonfinite scores are no longer accepted. - `memory_get` accepts offsets, document budgets and neighboring captures, and returns JSON text with continuation and provenance. Responses from an older core are bounded client-side and marked `paging: "client"`; neighboring captures require the updated core. - Search results include `doc_len` while retaining their existing excerpts. ### Validation - Added default-off negative cases for connection checks, ordinary edits, translation and changed decisions, plus opt-in checks for malformed results, superseded memories and workspace boundaries. These test retrieval policy, not embedding or answer accuracy. ## [0.2.0] - 2026-08-31 ### Changed - Autocontext requests `durable_only` search and filters client-side: injected cards only contain deliberate or consolidated memories. Raw transcript capture stays available through explicit `memory_search` calls. ## [0.1.0] - 2026-08-30 First public npm release. ### Added - Five canonical memory tools: `memory_remember`, `memory_search`, `memory_get`, `memory_update`, and `memory_delete`. - Four opt-in vault tools backed by the memnest core's encrypted secret store. - Workspace-scoped Autocontext for prompt-time semantic recall. - A `/memnest` command for service health, memory count, active data directory, and search latency. - A bundled Node.js 20 ESM extension with a six-file npm package. ### Behavior - Omitted projects use pi's absolute working directory so the core can isolate workspaces and include the shared `playbook` collection. - Autocontext searches substantive prompts in their original language and injects only results that pass its score threshold. - Retrieved text is labeled as untrusted reference data and stored markup is escaped before prompt injection. - Conversation capture remains in the core's `memnest watch` command, avoiding duplicate extension-side AutoLog hooks. - Search output omits the removed `recall_id` field instead of displaying an undefined value. ### Security - Vault tools are disabled unless `MEMNEST_EXPOSE_SECRET_TOOLS=1` is set before pi starts. - Sensitive values are rejected by `memory_remember` and must use the vault. - HTTP failures and unavailable vault crypto fail closed.