@snoai/memory 0.0.0-stage → 1.0.0

package/skills/extract-atomic-memory/references/resolve-subject.md

@@ -0,0 +1,19 @@
+## Resolve an unresolved subject
+
+You are shown claims whose subject was `unresolved`, and a list of candidates, each with its
+`display_name`. For each claim, pick the candidate whose display name is the thing the claim
+states a field of.
+
+A claim that says "the budget", "the proposal's deadline" or "a key point of the email" is a
+field of the document the conversation is editing — pick that document. A claim left unresolved
+is stored where no later revision can ever replace it, and the old and the new value are then
+both reported as current.
+
+A claim that adds, removes or changes a field of "the email", "the document", "the meeting" or
+"the proposal" is a field of the document the conversation is editing, and a candidate whose
+display name is that document's purpose or title sentence is the document even though the claim
+never repeats the name. Only when the turns show a second document of the same kind being
+started, and the claim belongs to it, is a candidate named earlier not the one: answer null then.
+
+Answer null only when no candidate is the thing: two candidates could equally be it, or the claim
+is about something none of them names. Answer with the candidate's id, never with a new name.

package/skills/extract-atomic-memory/references/surrounding-context.md

@@ -0,0 +1,10 @@
+## Surrounding context
+
+`preceding_context` and `following_context` contain earlier and later conversation turns,
+for understanding only. Resolve a short answer against its question and resolve references
+against what those turns actually say. Do not guess a new activity or object when the
+surrounding conversation names it.
+
+Extract and classify only the numbered `transcript`, not the surrounding context. A claim
+may use a name or relationship supplied by the context, but its source quote and turn index
+must come from the numbered transcript. Do not copy context-only claims into new records.

package/skills/extract-atomic-memory/references/user-subject-guard.md

@@ -0,0 +1,29 @@
+## User-subject guard
+
+You are shown a numbered batch of profile records. Decide each one separately: does the quoted
+user text support the claim as a fact about the user?
+
+Judge the meaning; the quote and the claim may be in different languages. A change the user
+reports about themselves ("I switched my editor to Emacs", "我把主力编辑器换成了 Emacs", "I moved
+to Austin") supports the state it leaves behind. An intention is a fact about the user too: what
+they say they need to do, plan to do, want, or look forward to supports a claim about that
+intention — "I also need to visit the university library sometime soon" supports "the user needs
+to visit the university library". Do not answer no because the thing has not happened yet. A
+liking, a taste, an opinion the user states as their own is a fact about the user whether or not
+a key was found for it; the presence or absence of an attribute is not evidence either way.
+
+Answer no when the quoted text is about another person, is the assistant's words, or does not
+mention the claim — "I switched to Emacs" does not say the user likes Emacs, and one question
+about a topic does not say the user is interested in it. A quote that cannot be read on its own
+("I really liked it") supports nothing. The subject proposal, the claim text and the attribute are
+not evidence by themselves; only the quote is.
+
+An explicit first-person preference ("My favorite drink is tea") directly supports the same
+preference stated about the user. No additional biography or evidence of repeated behavior is
+needed. This does not support a different preference that the quote never states.
+
+For this task, return only the `decisions` object in the supplied response schema, not the
+extraction `claims_found`/`records` envelope. Return one boolean decision for every supplied
+`record_index`, copying each index exactly once. Do not omit an index or return null: use true
+when the quote supports the claim and false when it does not. Check that every supplied index
+has one decision before returning.

package/skills/judge-group-closure/SKILL.md

@@ -0,0 +1,25 @@
+---
+name: judge-group-closure
+description: Judge which offered rows one nominated statement retires or replaces.
+---
+
+# Judge Group Closure
+
+## Group Closure
+
+- A nominated row is a retirement statement when it says that a fact in one candidate row is no
+  longer current, including when it describes that fact as a past preference.
+- Past tense or a shared topic is not enough. Choose a target only when the nominated row makes
+  the candidate row's present claim no longer true.
+- Return null when both rows can be true at the same time, including when they describe different
+  strengths of interest or different aspects of one topic.
+- Treat strength as part of the fact. A past strong preference does not retire a present weaker
+  interest in the same topic. It retires only a candidate that claims the same strong preference
+  is current.
+- Example: "used to be interested in X" retires "enjoys X" because it says the base interest is
+  no longer current.
+- Counterexample: "used to really like X" does not retire "is currently drawn to X" because a
+  former strong preference and a present weaker interest can both be true. Return null.
+- Choose every candidate row whose current meaning the nominated row retires.
+- Return those candidate row ids. Return an empty set when the nominated row retires no offered candidate.
+- Choose only by meaning. Shared words, text order, and age do not identify the target row.

package/skills/judge-rem-memory-updates/SKILL.md

@@ -0,0 +1,56 @@
+---
+name: judge-rem-memory-updates
+description: Judge REM memory relations, rewrites, verification, and clause carry by meaning. Use for REM update model calls that decide which older facts are superseded, which facts remain current, and whether a proposed rewrite preserves all surviving facts.
+---
+
+# Judge REM Memory Updates
+
+Judge meaning. Do not decide from shared words, punctuation, or text position.
+
+## Relation Judgment
+
+- Mark `supersedes` true only when the newer record replaces meaning in the older record.
+- Mark `retires_anything` true only when at least one meaning in the older record stops being current.
+- Mark `supersedes_everything` true only when every current meaning in the older record is replaced.
+
+## Source Rewrite
+
+- Put only the facts that remain current in `proposed_current`.
+- Quote each superseded value from the source row in `retired_values`.
+- Drop the wording that exists only to describe the change itself — what the value was
+  before, when it changed, and who changed it. That wording retires with the value it
+  describes. "currently $1,100,000, updated from $900,000 on 2025-06-06" rewrites to
+  "currently $1,100,000", and `retired_values` carries "$900,000".
+- This holds in every language and for every kind of value, not only amounts. Any phrasing of
+  the shape "changed from A to B", "shifted from A to B", "switched from A to B", "moved from A
+  to B", "A is now B", "no longer A but B" leaves ONLY B in `proposed_current` and puts A in
+  `retired_values`. The old value must not appear anywhere in the rewrite, not even inside the
+  phrase that describes the change — copying "shifted from A to B" into `proposed_current`
+  keeps the retired value alive and is wrong. Rewrite it as B alone.
+  English: "shifted from morning tea to coffee" rewrites to "coffee", retiring "morning tea".
+  Spanish: "cambió de té matutino a café" rewrites to "café", retiring "té matutino".
+- A line that says an item is no longer needed, no longer tracked, done with, or that asks
+  for it to be removed from the list, IS a retirement instruction. Drop that whole line from
+  `proposed_current`, and drop the item it names as well. Quote the named item in
+  `retired_values`. Leaving either one in place is wrong: the instruction line is wording
+  about a change, and the item it names is the value that change retires.
+- Return no retired values when nothing was superseded.
+- Return an empty proposed value when nothing remains current.
+
+## Rewrite Verification
+
+- Mark `faithful` true only when source evidence supports every claim in the rewrite.
+- When a newer record is provided, retire only meanings that the newer record replaces.
+- Reject any unsupported fact, entity, quantity, or claim.
+- Mark `retired_absent` true only when no retired meaning remains current in the rewrite.
+- Mark `all_facts_accounted` true only when every source fact remains current or appears in the retired values.
+- Wording that exists only to describe the change — the prior value, the date it changed,
+  who changed it — is part of the retirement, not a surviving fact. Its absence from the
+  rewrite never makes `all_facts_accounted` false.
+- Accept paraphrases that keep the same meaning.
+
+## Clause Carry
+
+- Mark a clause already current only when the survivor asserts the same meaning as current.
+- A negation, quotation, or past description is not a current assertion.
+- Different words with the same meaning count as current.

package/skills/judge-rem-memory-updates/agents/openai.yaml

@@ -0,0 +1,4 @@
+interface:
+  display_name: "Judge REM memory updates"
+  short_description: "Judge which memory facts stay current"
+  default_prompt: "Judge the supplied REM memory update by meaning and return the requested JSON contract."

package/skills/judge-retired-position/SKILL.md

@@ -0,0 +1,25 @@
+---
+name: judge-retired-position
+description: Judge whether one stored row states a retired profile position as current.
+---
+
+# Judge Retired Position
+
+## Current Position Match
+
+- Answer yes only when the candidate row states, as current, the same position that the profile lifecycle retired.
+- A simple present statement such as "The user likes X" states a current position. It does not need the word "now".
+- When the candidate repeats that position with the same meaning, answer yes.
+- A dated occurrence is history. Answer no even when it mentions the same subject or words.
+- A plan, quotation, comparison, question, or third-party statement is not the user's current position.
+- Surface similarity is not enough. Judge the meaning of the whole row.
+- When the evidence does not settle the meaning, answer no.
+
+## Retirement Recheck
+
+- Judge one retirement inside a current-state profile section.
+- A stored clause was retired when the incoming assertion arrived. Decide whether that retirement is right.
+- Answer retire=true only when the incoming assertion states a newer value of the same fact, removes it, makes it historical, or restates this same fact in other words so that keeping both would say one thing twice.
+- For section preferences.general only: retire=true is also right when a listed live sibling section already states this same fact in its content.
+- A clause about a different subject stays. A clause the incoming assertion does not mention stays, even when it is the same broad topic.
+- When the evidence does not settle it, answer retire=false.

package/skills/judge-task-lifecycle/SKILL.md

@@ -0,0 +1,45 @@
+---
+name: judge-task-lifecycle
+description: Judge how a memory changes active-task state and whether it refers to an existing task. Use for mem-claw task-lifecycle model calls.
+---
+
+# Judge Task Lifecycle
+
+Judge task meaning. Do not decide from exact wording alone.
+
+## Active Task State
+
+Apply the `To-do Boundary` section supplied from `extract-atomic-memory` before these state rules.
+
+Decide in this order and stop at the first step that applies.
+
+1. **Does the candidate report OR INSTRUCT that an existing active task is done, finished,
+   handled, cancelled, or dropped?** If yes, the answer is `complete` or `remove` on THAT task.
+   Never `open_or_refine`, and never `none`.
+   A report ("I finished X", "X is done") and an instruction ("mark X complete", "tick X off",
+   "close X", "remove X from my list") both land here. An instruction to change a task's state
+   IS the state change; do not treat it as a request you decline to act on.
+   A candidate that speaks about an active task's ending is never a new task, however it is worded,
+   and restating the task's name inside such a sentence is naming the task to close, not opening one.
+2. Otherwise, if the candidate describes work still to be done, use `open_or_refine`.
+3. Otherwise use `none`.
+
+- Use `open_or_refine` only for a current task that remains active.
+- Use `complete` only when the candidate says an active task was achieved.
+- Use `remove` only when the candidate says an active task was cancelled or should no longer be tracked.
+- An explicit instruction to mark a task complete, done, or finished is `complete`, even when it is
+  phrased as an edit to a list — "mark X complete on my todo list" and "tick X off" are `complete`,
+  not `remove`. Taking a finished task off a list is what completing it looks like; the words that
+  decide are whether the work was done. Reserve `remove` for a task abandoned, cancelled, called off,
+  no longer needed, or added by mistake — cases where the work will NOT happen.
+- Use `none` when the candidate does not change active-task state.
+- A candidate that reports a task as finished, cancelled, or already handled never opens one.
+- If that task is active, complete or remove it. If it is not active, make no task change.
+- Match tasks by meaning, not exact wording. Refine an existing task instead of opening a duplicate.
+
+## Existing Task Relation
+
+- Use `same_instance` when the assertion refers to one listed active task, including different wording with the same meaning.
+- Use `distinct_instance` when the assertion is a different active task.
+- Use `none` when the assertion does not describe an active-task relation.
+- Use `uncertain` when the evidence does not settle the relation.

package/skills/key-state-attribute/SKILL.md

@@ -0,0 +1,16 @@
+---
+name: key-state-attribute
+description: Choose one offered state attribute for a stored statement about a named thing.
+---
+
+# Key State Attribute
+
+## State Attribute
+
+- Read the complete statement before you choose.
+- Choose the offered slug whose meaning best describes the fact in the statement.
+- Choose only one slug from the offered list.
+- Return `null` when no offered slug describes the fact.
+- Return `null` when you are not certain.
+- Do not choose from shared words alone.
+- Return only the slug or `null`. Do not add prose.

package/skills/resolve-entity-identity/SKILL.md

@@ -0,0 +1,24 @@
+---
+name: resolve-entity-identity
+description: Resolve a new entity display name to an existing entity id or to a new entity.
+---
+
+# Resolve Entity Identity
+
+## Entity Identity
+
+- You are shown one new display name and the project's existing entity display names.
+- Answer with an existing entity id only when the new display name identifies that same entity.
+- The same entity is very often written more than one way. A shortening ("ACME Rebrand" for
+  "Acme Corp Rebrand"), a change of case or spacing, an added or dropped type word ("… project",
+  "… proposal", "… email"), an added or dropped honorific, or the name wrapped in quotes all
+  name the SAME entity. Answer that entity's id.
+- Otherwise answer `new`.
+- Answer `new` when you are in doubt whether two different things are meant — two people who
+  share a surname, two documents with similar titles. A different spelling of one name is not
+  doubt.
+- A wrong merge silently fuses the facts of two people or two documents.
+- A wrong split leaves two current values for a thing that can only have one — two budgets for
+  one project — and both are served as true.
+- Return exactly one JSON object and no prose: `{ "entity_id": "new" }`.
+- Set `entity_id` to `new` or to one entity id shown with the existing display names.

package/skills/resolve-relative-time/SKILL.md

@@ -0,0 +1,22 @@
+---
+name: resolve-relative-time
+description: Understand temporal meaning in context without performing calendar arithmetic.
+---
+
+# Understand Relative Time
+
+Read the complete context to determine which event a time expression describes and whether
+it is past, future, ongoing, quoted, or uncertain. Preserve the source words and its precision.
+Never calculate a replacement date. The shared date resolver asks for a structured calendar
+instruction and code performs that instruction. Keywords alone cannot settle temporal meaning.
+
+## With Session Anchor
+
+Use the session as a reference only when the speaker refers to it. Distinguish a duration from
+an event date and a current report from an event happening now. The same words can describe
+different times in different contexts. Preserve the fact when its date remains unknown.
+
+## Without Session Anchor
+
+Keep relative expressions unresolved when their reference is absent. Preserve explicit date
+components that the source states. Do not supply the current year or day from your own clock.

package/sqlite-extensions/LICENSE-simple-tokenizer

@@ -0,0 +1,28 @@
+Vendored binaries in this directory are built from
+https://github.com/wangfenjin/simple (release v0.7.1).
+
+Upstream is dual-licensed MIT/GPLv3+ at the user's choice. We redistribute
+under the MIT terms reproduced below.
+
+----------------------------------------------------------------------
+Copyright (c) 2020, 2025 Wang Fenjin
+
+MIT License
+
+Permission is hereby granted, free of charge, to any person obtaining a copy
+of this software and associated documentation files (the "Software"), to deal
+in the Software without restriction, including without limitation the rights
+to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+copies of the Software, and to permit persons to whom the Software is
+furnished to do so, subject to the following conditions:
+
+The above copyright notice and this permission notice shall be included in all
+copies or substantial portions of the Software.
+
+THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+SOFTWARE.

package/sqlite-extensions/LICENSE-sqlite-vec

@@ -0,0 +1,21 @@
+MIT License
+
+Copyright (c) 2024 Alex Garcia
+
+Permission is hereby granted, free of charge, to any person obtaining a copy
+of this software and associated documentation files (the "Software"), to deal
+in the Software without restriction, including without limitation the rights
+to use, copy, modify, merge, publish, distribute, sublicense, and/or sell
+copies of the Software, and to permit persons to whom the Software is
+furnished to do so, subject to the following conditions:
+
+The above copyright notice and this permission notice shall be included in all
+copies or substantial portions of the Software.
+
+THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS OR
+IMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,
+FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THE
+AUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHER
+LIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,
+OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THE
+SOFTWARE.