Skip to content
verify mcp Beta VerifyMCP is currently in beta. If you notice any issues, get in touch and we’ll put it right.

io.github.Mibayy/token-savior

PYPI · TOKEN-SAVIOR-RECALL · SCANNED SEP 21

Structural codebase MCP server with persistent memory: navigate by symbol, recall across sessions.

−15 this week 66 Trust /100
Trust breakdown (7 categories)

How this component scores in each security and reliability category. Every signal is checked automatically from public evidence about the published package, including repeated runs of it in an isolated sandbox, and we only credit what we can confirm. How we score → Why this is hard to score →

Supply Chain Security50
  • Malware scan not yet available for this package.Unverified
  • No known CVEs affecting this package version or its production dependencies.Pass
  • Runs hatchling.build at install time, a recognised native-build step with no shell scripting around it. View diagnostics → Pass
  • 0 of 34 dependencies flagged as unhealthy. View diagnostics → Pass
Provenance & Transparency45
Schema Quality & AI Usability75
  • AI-judged instruction clarity (good).Pass
  • Tool/resource definitions use about 6320 tokens (~91/item across 69 items; 69 tools + 0 resources), lean.Pass
  • Usage-examples check failed: none of the tools include examples. See how to fix → Fail
Stability & Change Management87
  • Stability observed for 26 of 30 days with no destabilising changes; credit accrues until the full window elapses.Partial
Tool Coverage94
  • 100% of tools have a non-trivial description (not blank, and not just the tool's name).Pass
  • 81% of tool parameters carry a description.Partial
Tool Safety75
  • No prompt-injection markers were found in the server instructions, tool names or descriptions we captured.Pass
  • 0 of 3 tool(s) whose name or description implies an irreversible operation declare an MCP destructiveHint annotation; "memory_delete" implies "delete" and declares no destructiveHint at all, which the MCP spec reads as destructive by default. See how to fix → Fail
  • An AI judge read all 69 captured unit(s) of tool text and found none that tries to manipulate the model reading it.Pass
Capabilities100
  • Implements a supported MCP spec version (2025-11-25); the latest is 2026-07-28.Pass
Install

How do I install the io.github.Mibayy/token-savior MCP server?

io.github.Mibayy/token-savior runs locally as a PyPI package, launched with uvx token-savior-recall. Ready-made configuration for Claude, Cursor, VS Code, Codex and 5 more is on this page, copied from each client's own documentation.

pypi · token-savior-recall

# add to Claude Code
claude mcp add mibayy-token-savior -- uvx token-savior-recall
// .cursor/mcp.json
{
  "mcpServers": {
    "mibayy-token-savior": {
      "command": "uvx",
      "args": [
        "token-savior-recall"
      ]
    }
  }
}
// .vscode/mcp.json
{
  "servers": {
    "mibayy-token-savior": {
      "command": "uvx",
      "args": [
        "token-savior-recall"
      ]
    }
  }
}
# add to Codex CLI
codex mcp add mibayy-token-savior -- uvx token-savior-recall
// opencode.json
{
  "$schema": "https://opencode.ai/config.json",
  "mcp": {
    "mibayy-token-savior": {
      "type": "local",
      "command": [
        "uvx",
        "token-savior-recall"
      ],
      "enabled": true
    }
  }
}
# add to OpenClaw
openclaw mcp add mibayy-token-savior --command uvx --arg token-savior-recall
# ~/.hermes/config.yaml
mcp_servers:
  mibayy-token-savior:
    command: "uvx"
    args: ["token-savior-recall"]
// ~/.netclaw/config/netclaw.json
{
  "McpServers": {
    "mibayy-token-savior": {
      "Transport": "stdio",
      "Command": "uvx",
      "Arguments": [
        "token-savior-recall"
      ]
    }
  }
}
# add to Vellum
assistant mcp add mibayy-token-savior -t stdio -c uvx -a token-savior-recall
// mcp.json
{
  "mcpServers": {
    "mibayy-token-savior": {
      "command": "uvx",
      "args": [
        "token-savior-recall"
      ]
    }
  }
}
Changelog

Every change we have recorded for this component, newest first. Security-relevant changes are always shown. ▲ marks a change for the better, ▼ a change for the worse; unmarked changes are neutral.

  • 21 Sept 26 +1

    No change was recorded against any check on this day. Stability & Change Management went from 83 to 87. That category is still filling its 30-day observation window: 25 days of observed history at the previous scan, 26 at this one. The score rises as the window fills, whether or not the server changes.

  • 19 Sept 26 −3
    • Stability: pass → 0.80 functional
  • 18 Sept 26 +1
    • Stability: 0.97 → pass security
  • 17 Sept 26 −15
    • Malware scan: pass → unverified security
  • 16 Sept 26 +16
    • Malware scan: unverified → pass security
  • 15 Sept 26 −15
    • Malware scan: pass → unverified security
  • 14 Sept 26 +16
    • Malware scan: unverified → pass security
  • 13 Sept 26 −15
    • Malware scan: pass → unverified security
Diagnostics

Diagnostic detail from the automated scan of this channel: what the scanner observed at each step, so you can see exactly where a check passed or failed. It is informational only and never changes the trust score.

Captured 21 Sept 2026 · Analysed pypi/token-savior-recall@4.13.0

Provenance No attestation

The registry publishes no build provenance for this version, so there is nothing to verify.

Result No attestation
Ecosystem pypi

Background: How many MCP packages publish verified provenance →

Install scripts 1 script
Hook Tier Command
build_backend allowlisted hatchling.build

Background: Why install scripts are a supply-chain risk →

Dependencies 34 packages
Packages resolved 34
Tree resolution Complete

Background: SBOMs and build attestations, explained →

MCP tools · 69 exposed · ~6,320 tokens

The tools this component advertises to a client, with an estimated token cost for each. Expand a tool to see its parameters and schema. The per-tool counts are indicative and are not scored directly; the schema's total context footprint is one signal in Schema Quality & AI Usability. A tool's description is untrusted text the model reads on every call, which is what makes this list a security surface and not just an inventory: how tool poisoning works →

Tool Tokens
memory_get ~101

Layer 3: full observation content by IDs (~200 tokens/result). Final progressive-disclosure layer.

NameTypeReqDescription
fullbooleanIf false (default), content trimmed to 80 chars. If true, full content.
idsarrayyesObservation IDs. Each item may be an integer (42), a digit string ("42"), or a citation URI ("ts://obs/42").
projectstringProject name/path (default: active).

No output schema declared.

No examples provided.

memory_index ~74

Layer 1: compact index of recent observations — ID, type, title, importance, age, citation URI.

NameTypeReqDescription
limitintegerMax entries to return (default 30).
projectstringProject name/path (default: active).
type_filterstringFilter by observation type (optional).

No output schema declared.

No examples provided.

memory_save ~183

Persist a fact, guardrail, or note across sessions.

NameTypeReqDescription
conceptsstringOptional conceptual tags (JSON array or comma list).
contentstringyes
contextstring
factsstringOptional atomic facts (JSON array or bullet list).
file_pathstring
how_to_applystring
importanceinteger1-10
is_globalboolean
narrativestringOptional free-form narrative explaining the obs in prose.
projectstringProject name/path (default: active).
session_idinteger
symbolstring
tagsarray
titlestringyes
ttl_daysinteger
typestringyes
whystring

No output schema declared.

No examples provided.

memory_search ~78

Layer 2 FTS5 search over memory observations, compact rows with snippets (~60 tokens/result).

NameTypeReqDescription
limitintegerDefault 20.
projectstringProject name/path (default: active).
querystringyesFTS5 (AND/OR/NOT/phrase).
type_filterstring

No output schema declared.

No examples provided.

move_symbol ~81

Move a symbol to a different file, updating imports in all call sites.

NameTypeReqDescription
create_if_missingbooleanCreate target file if it doesn't exist (default true).
projectstringProject name/path (default: active).
symbolstringyesSymbol name to move.
target_filestringyesRelative path to the target file.

No output schema declared.

No examples provided.

reasoning_list ~46

List stored reasoning chains by access_count then recency.

NameTypeReqDescription
limitintegerMax rows (default 50).
projectstringProject name/path (default: active).

No output schema declared.

No examples provided.

reasoning_save ~104

Persist a reasoning trace (goal + steps + conclusion) for later reuse.

NameTypeReqDescription
conclusionstringyes
confidencenumber0.0-1.0 (default 0.8).
evidence_obs_idsarray
goalstringyes
projectstringProject name/path (default: active).
stepsarrayyes[{tool,args,observation},...]
ttl_daysinteger

No output schema declared.

No examples provided.

reasoning_search ~80

Search stored reasoning chains by goal similarity (FTS5 + Jaccard).

NameTypeReqDescription
limitintegerMax rows (default 5).
projectstringProject name/path (default: active).
querystringyesGoal-like query text.
thresholdnumberMinimum Jaccard similarity (default 0.3).

No output schema declared.

No examples provided.

reindex ~44

Rebuild the project index.

NameTypeReqDescription
forcebooleanRebuild even if no mtime changes detected.
projectstringProject name/path (default: active).

No output schema declared.

No examples provided.

replace_symbol_source ~81

Replace an indexed symbol's full source block directly.

NameTypeReqDescription
file_pathstringOptional file path to disambiguate symbols.
new_sourcestringyesReplacement source for the symbol.
projectstringProject name/path (default: active).
symbol_namestringyesFunction, method, class, or section name to replace.

No output schema declared.

No examples provided.

run_impacted_tests ~92

Run pytest on files impacted by the current worktree changes.

NameTypeReqDescription
changed_filesarray
compactboolean
include_outputboolean
max_output_charsinteger
max_testsinteger
projectstringProject name/path (default: active).
symbol_namesarray
timeout_secinteger

No output schema declared.

No examples provided.

run_project_action ~87

Run a discovered project action by id (bounded output, bounded timeout).

NameTypeReqDescription
action_idstringyese.g. 'python:test', 'npm:test'.
include_outputboolean
max_output_charsintegerDefault 12000.
projectstringProject name/path (default: active).
timeout_secintegerDefault 120.

No output schema declared.

No examples provided.

search_codebase ~140

Regex (default) or semantic (semantic=true) search across indexed files.

NameTypeReqDescription
ignore_generatedbooleanSkip generated/minified files (default true). Regex mode only.
max_resultsintegerMaximum number of results to return (default 100, 0 = unlimited).
patternstringyesRegex pattern (regex mode) or natural-language description (semantic mode).
projectstringProject name/path (default: active).
semanticbooleanIf true, interpret `pattern` as a description and rank symbols by embedding cosine similarity. Returns enriched hits with signature/docstring/score. Default false (regex).

No output schema declared.

No examples provided.

search_in_symbols ~78

Regex search that returns the enclosing function/class for each match, in addition to file:line.

NameTypeReqDescription
max_resultsintegerMaximum number of results to return (default 100, 0 = unlimited).
patternstringyesRegular expression pattern to search for.
projectstringProject name/path (default: active).

No output schema declared.

No examples provided.

set_project_root ~76

Register a new project root and switch to it. If the path is already registered, this becomes a cheap active-root switch (no reindex) unless force=true is set.

NameTypeReqDescription
forcebooleanRebuild the index even if the project is already registered.
pathstringyesAbsolute path to the project root directory.

No output schema declared.

No examples provided.

switch_project ~30

Switch the active project.

NameTypeReqDescription
namestringyesProject name (basename of path) or full path.

No output schema declared.

No examples provided.

ts_discover ~221

Scan Claude Code transcripts for missed Token Savior opportunities (native-call chains that should have been a single TS call) and/or report TS-vs-native adoption ratios. Scans ALL transcript projects by default; pass project=<substring> to filter. format='table' (default) or 'json' returns ranked Findings; format='adoption' or 'adoption_json' returns the TS adoption summary with trend + worst sessions. Read-only, PII-safe.

NameTypeReqDescription
formatstringOutput format. 'table' (default) / 'json' = ranked missed-opportunity Findings. 'adoption' / 'adoption_json' = TS vs native adoption summary.
limitintegerCap the number of findings returned (table/json only; default unlimited).
projectstringFilter to sanitized project dirs whose name contains this substring (e.g. '-root'). Omit to scan ALL transcript projects under ~/.claude/projects/.
since_daysintegerOnly consider events newer than now - since_days (default 7).

No output schema declared.

No examples provided.

ts_execute ~131

Run a JS script in a Node sandbox with a typed facade. The script body executes as `async () => { <body> }`; use `await tools.<name>(args)` to call any allowed tool (find_symbol, get_function_source, get_dependents, search_codebase, replace_symbol_source, etc.). Return the final value. Collapses find->read->deps chains into one round-trip.

NameTypeReqDescription
scriptstringyesJS function body. Use `await tools.foo(args)` then `return value`.
timeout_msintegerMax script wall-clock in ms (default 30000).

No output schema declared.

No examples provided.

ts_search ~190

Find the top-K Token Savior tools most relevant to a natural-language query via embedding cosine similarity. Use when the manifest is in 'tiny' or 'code_mode' profile or you don't know which tool fits. Returns each candidate with its full inputSchema (format='schema') or a one-line TypeScript signature (format='ts') suitable for ts_execute scripts.

NameTypeReqDescription
formatstring'schema' (default) returns JSONSchema; 'ts' returns a TypeScript signature for Code Mode scripts. Auto-set to 'ts' when profile=code_mode.
include_schemabooleanIf false, drop inputSchema from the response (smaller payload).
querystringyesNatural-language description of what you need (e.g. 'find dependents of update_user').
top_kintegerHow many tool candidates to return (default 5, max 12).

No output schema declared.

No examples provided.

Common questions

What is the io.github.Mibayy/token-savior MCP server?

io.github.Mibayy/token-savior is an MCP server listed in the public MCP registry as io.github.Mibayy/token-savior. Structural codebase MCP server with persistent memory: navigate by symbol, recall across sessions. This page covers its PyPI package (token-savior-recall).

Is the io.github.Mibayy/token-savior MCP server safe to use?

io.github.Mibayy/token-savior scores 66 out of 100 on VerifyMCP. We found no known CVEs affecting it as of 21 September 2026. That is a record of what we were able to check automatically, not an endorsement. The category breakdown on this page shows every signal behind the number, including the ones we could not confirm.

What tools does the io.github.Mibayy/token-savior MCP server expose?

io.github.Mibayy/token-savior exposes 69 tools: list_projects, switch_project, get_git_status, get_changed_symbols, build_commit_summary, and 64 more. Their descriptions and schemas cost roughly 6,320 tokens of context every time the server is loaded.

Is the io.github.Mibayy/token-savior MCP server still maintained?

io.github.Mibayy/token-savior is still listed as active in the MCP registry. We last reached this channel on 21 September 2026. Those dates come from our own scans of the registry and the channel itself, not from anything the publisher announced.

What licence is the io.github.Mibayy/token-savior MCP server under?

io.github.Mibayy/token-savior declares the MIT licence, which is OSI-approved. That covers the source only, and says nothing about the cost of any service it calls.